测试结果：

#PAN My

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
# python test_models.py something --VAP --batch_size=16 -j=4 --test_crops=1 --test_segments=8,8,8,8 --weights=pretrained/PAN_RGB_something_resnet50_shift8_blockres_avg_segment8_e50.pth.tar,pretrained/PAN_RPA_something_resnet50_shift8_blockres_avg_segment8_e80.pth.tar,pretrained/PAN_PA_something_resnet50_shift8_blockres_avg_segment8_e80.pth.tar,pretrained/PAN_RGB_something_resnet50_shift8_blockres_avg_segment8_e50.pth.tar
$ bash scripts/test/sthv1/AAA.sh
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     RGB
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256

=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
video number:11522
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     PA
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256

=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
video number:11522
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     PA
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256

=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
video number:11522
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     RGB
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256

=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
video number:11522
video 0 done, total 0/11522, average 0.223 sec/video, moving Prec@1 50.000 Prec@5 81.250
video 320 done, total 320/11522, average 0.055 sec/video, moving Prec@1 53.869 Prec@5 83.631
video 640 done, total 640/11522, average 0.051 sec/video, moving Prec@1 54.268 Prec@5 82.165
video 960 done, total 960/11522, average 0.049 sec/video, moving Prec@1 53.484 Prec@5 80.943
video 1280 done, total 1280/11522, average 0.049 sec/video, moving Prec@1 53.627 Prec@5 81.481
video 1600 done, total 1600/11522, average 0.048 sec/video, moving Prec@1 53.713 Prec@5 80.941
video 1920 done, total 1920/11522, average 0.048 sec/video, moving Prec@1 52.789 Prec@5 80.527
video 2240 done, total 2240/11522, average 0.048 sec/video, moving Prec@1 52.527 Prec@5 80.452
video 2560 done, total 2560/11522, average 0.048 sec/video, moving Prec@1 52.446 Prec@5 80.551
video 2880 done, total 2880/11522, average 0.048 sec/video, moving Prec@1 52.106 Prec@5 80.352
video 3200 done, total 3200/11522, average 0.048 sec/video, moving Prec@1 51.710 Prec@5 80.379
video 3520 done, total 3520/11522, average 0.047 sec/video, moving Prec@1 51.669 Prec@5 80.402
video 3840 done, total 3840/11522, average 0.047 sec/video, moving Prec@1 51.504 Prec@5 80.031
video 4160 done, total 4160/11522, average 0.047 sec/video, moving Prec@1 51.245 Prec@5 79.957
video 4480 done, total 4480/11522, average 0.047 sec/video, moving Prec@1 51.157 Prec@5 80.004
video 4800 done, total 4800/11522, average 0.047 sec/video, moving Prec@1 51.017 Prec@5 79.672
video 5120 done, total 5120/11522, average 0.047 sec/video, moving Prec@1 50.818 Prec@5 79.420
video 5440 done, total 5440/11522, average 0.047 sec/video, moving Prec@1 50.953 Prec@5 79.582
video 5760 done, total 5760/11522, average 0.047 sec/video, moving Prec@1 50.952 Prec@5 79.415
video 6080 done, total 6080/11522, average 0.047 sec/video, moving Prec@1 50.968 Prec@5 79.560
video 6400 done, total 6400/11522, average 0.047 sec/video, moving Prec@1 51.044 Prec@5 79.723
video 6720 done, total 6720/11522, average 0.047 sec/video, moving Prec@1 50.920 Prec@5 79.751
video 7040 done, total 7040/11522, average 0.047 sec/video, moving Prec@1 51.020 Prec@5 79.691
video 7360 done, total 7360/11522, average 0.047 sec/video, moving Prec@1 50.976 Prec@5 79.542
video 7680 done, total 7680/11522, average 0.047 sec/video, moving Prec@1 51.130 Prec@5 79.691
video 8000 done, total 8000/11522, average 0.047 sec/video, moving Prec@1 51.085 Prec@5 79.753
video 8320 done, total 8320/11522, average 0.047 sec/video, moving Prec@1 51.116 Prec@5 79.762
video 8640 done, total 8640/11522, average 0.047 sec/video, moving Prec@1 50.947 Prec@5 79.737
video 8960 done, total 8960/11522, average 0.047 sec/video, moving Prec@1 50.925 Prec@5 79.768
video 9280 done, total 9280/11522, average 0.047 sec/video, moving Prec@1 50.968 Prec@5 79.733
video 9600 done, total 9600/11522, average 0.047 sec/video, moving Prec@1 51.113 Prec@5 79.773
video 9920 done, total 9920/11522, average 0.047 sec/video, moving Prec@1 51.127 Prec@5 79.781
video 10240 done, total 10240/11522, average 0.047 sec/video, moving Prec@1 51.268 Prec@5 79.904
video 10560 done, total 10560/11522, average 0.047 sec/video, moving Prec@1 51.314 Prec@5 79.870
video 10880 done, total 10880/11522, average 0.047 sec/video, moving Prec@1 51.386 Prec@5 79.947
video 11200 done, total 11200/11522, average 0.047 sec/video, moving Prec@1 51.444 Prec@5 79.913
video 11520 done, total 11520/11522, average 0.047 sec/video, moving Prec@1 51.311 Prec@5 79.865
[0.88059701 0.2875     0.25714286 0.64179104 0.39473684 0.42241379
 0.71641791 0.41176471 0.65048544 0.70873786 0.38666667 0.44444444
 0.32978723 0.4516129  0.71974522 0.5        0.23493976 0.25
 0.44715447 0.36       0.336      0.5862069  0.48148148 0.41176471
 0.5483871  0.41666667 0.43478261 0.48       0.72727273 0.63492063
 0.59090909 0.57692308 0.83333333 0.23076923 0.32608696 0.13888889
 0.81609195 0.79411765 0.10526316 0.59259259 0.68595041 0.66
 0.73076923 0.67346939 0.71153846 0.59016393 0.41176471 0.42105263
 0.17977528 0.69924812 0.62264151 0.         0.36       0.4
 0.23809524 0.29166667 0.42424242 0.53125    0.         0.75510204
 0.78787879 0.16666667 0.66       0.19090909 0.1        0.44444444
 0.18518519 0.24242424 0.50704225 0.42307692 0.58064516 0.25
 0.55263158 0.52       0.78333333 0.48387097 0.35294118 0.52941176
 0.36363636 0.80487805 0.08571429 0.19047619 0.36666667 0.15789474
 0.65714286 0.39655172 0.59459459 0.63492063 0.16666667 0.08928571
 0.62162162 0.7804878  0.53488372 0.59259259 0.71028037 0.27419355
 0.03703704 0.11904762 0.56140351 0.47058824 0.32653061 0.55555556
 0.203125   0.59183673 0.63414634 0.65714286 0.52808989 0.72463768
 0.49122807 0.32967033 0.64705882 0.05882353 0.66216216 0.28571429
 0.61320755 0.4        0.34545455 0.76595745 0.31707317 0.58333333
 0.80681818 0.60784314 0.36363636 0.5        0.33333333 0.46938776
 0.49056604 0.57407407 0.11724138 0.625      0.28       0.25925926
 0.26315789 0.33333333 0.68211921 0.43103448 0.07142857 0.38333333
 0.15254237 0.59210526 0.66       0.37209302 0.40909091 0.48305085
 0.62666667 0.45774648 0.63207547 0.1641791  0.546875   0.832
 0.67878788 0.27419355 0.33333333 0.81481481 0.53571429 0.25490196
 0.27868852 0.3        0.51485149 0.34375    0.46296296 0.375
 0.3        0.37931034 0.7254902  0.75757576 0.9223301  0.87735849
 0.76851852 0.72972973 0.43809524 0.64748201 0.69343066 0.72368421]
upper bound: 0.4892586659876352
-----Evaluation is finished------
Class Accuracy 47.29%
Overall Prec@1 51.31% Prec@5 79.86%
/home/hs/anaconda3/envs/dzl/lib/python3.7/site-packages/numpy/core/_asarray.py:136: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray
  return array(a, dtype, copy=False, order=order, subok=True)

#PAN RGB&RPA

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
# python test_models.py something --VAP --batch_size=16 -j=4 --test_crops=1 --test_segments=8,8 --weights=pretrained/PAN_RGB_something_resnet50_shift8_blockres_avg_segment8_e50.pth.tar,pretrained/PAN_RPA_something_resnet50_shift8_blockres_avg_segment8_e80.pth.tar
$ bash scripts/test/sthv1/Full.sh
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     RGB
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256

=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
video number:11522
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     PA
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256

=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
video number:11522
video 0 done, total 0/11522, average 0.230 sec/video, moving Prec@1 56.250 Prec@5 81.250
video 320 done, total 320/11522, average 0.034 sec/video, moving Prec@1 55.357 Prec@5 84.226
video 640 done, total 640/11522, average 0.029 sec/video, moving Prec@1 55.488 Prec@5 82.317
video 960 done, total 960/11522, average 0.027 sec/video, moving Prec@1 53.689 Prec@5 80.225
video 1280 done, total 1280/11522, average 0.026 sec/video, moving Prec@1 54.090 Prec@5 80.478
video 1600 done, total 1600/11522, average 0.026 sec/video, moving Prec@1 54.084 Prec@5 79.950
video 1920 done, total 1920/11522, average 0.025 sec/video, moving Prec@1 53.202 Prec@5 79.649
video 2240 done, total 2240/11522, average 0.025 sec/video, moving Prec@1 52.660 Prec@5 79.699
video 2560 done, total 2560/11522, average 0.025 sec/video, moving Prec@1 52.484 Prec@5 79.852
video 2880 done, total 2880/11522, average 0.025 sec/video, moving Prec@1 52.072 Prec@5 79.593
video 3200 done, total 3200/11522, average 0.025 sec/video, moving Prec@1 51.617 Prec@5 79.602
video 3520 done, total 3520/11522, average 0.025 sec/video, moving Prec@1 51.499 Prec@5 79.638
video 3840 done, total 3840/11522, average 0.025 sec/video, moving Prec@1 51.167 Prec@5 79.305
video 4160 done, total 4160/11522, average 0.025 sec/video, moving Prec@1 50.766 Prec@5 79.239
video 4480 done, total 4480/11522, average 0.025 sec/video, moving Prec@1 50.512 Prec@5 79.115
video 4800 done, total 4800/11522, average 0.025 sec/video, moving Prec@1 50.353 Prec@5 78.738
video 5120 done, total 5120/11522, average 0.025 sec/video, moving Prec@1 50.097 Prec@5 78.466
video 5440 done, total 5440/11522, average 0.026 sec/video, moving Prec@1 50.220 Prec@5 78.666
video 5760 done, total 5760/11522, average 0.026 sec/video, moving Prec@1 50.208 Prec@5 78.497
video 6080 done, total 6080/11522, average 0.026 sec/video, moving Prec@1 50.213 Prec@5 78.658
video 6400 done, total 6400/11522, average 0.026 sec/video, moving Prec@1 50.374 Prec@5 78.787
video 6720 done, total 6720/11522, average 0.026 sec/video, moving Prec@1 50.282 Prec@5 78.741
video 7040 done, total 7040/11522, average 0.026 sec/video, moving Prec@1 50.425 Prec@5 78.699
video 7360 done, total 7360/11522, average 0.026 sec/video, moving Prec@1 50.298 Prec@5 78.552
video 7680 done, total 7680/11522, average 0.026 sec/video, moving Prec@1 50.468 Prec@5 78.768
video 8000 done, total 8000/11522, average 0.026 sec/video, moving Prec@1 50.424 Prec@5 78.780
video 8320 done, total 8320/11522, average 0.026 sec/video, moving Prec@1 50.468 Prec@5 78.791
video 8640 done, total 8640/11522, average 0.026 sec/video, moving Prec@1 50.347 Prec@5 78.789
video 8960 done, total 8960/11522, average 0.026 sec/video, moving Prec@1 50.379 Prec@5 78.877
video 9280 done, total 9280/11522, average 0.026 sec/video, moving Prec@1 50.398 Prec@5 78.873
video 9600 done, total 9600/11522, average 0.026 sec/video, moving Prec@1 50.520 Prec@5 78.900
video 9920 done, total 9920/11522, average 0.026 sec/video, moving Prec@1 50.533 Prec@5 78.925
video 10240 done, total 10240/11522, average 0.026 sec/video, moving Prec@1 50.575 Prec@5 79.027
video 10560 done, total 10560/11522, average 0.026 sec/video, moving Prec@1 50.643 Prec@5 78.971
video 10880 done, total 10880/11522, average 0.026 sec/video, moving Prec@1 50.734 Prec@5 79.102
video 11200 done, total 11200/11522, average 0.026 sec/video, moving Prec@1 50.785 Prec@5 79.110
video 11520 done, total 11520/11522, average 0.026 sec/video, moving Prec@1 50.720 Prec@5 79.040
[0.89552239 0.325      0.27619048 0.59701493 0.36842105 0.43103448
 0.68656716 0.41176471 0.67961165 0.67961165 0.36       0.45833333
 0.30851064 0.4516129  0.68789809 0.48       0.22891566 0.25
 0.45528455 0.34666667 0.328      0.53448276 0.42592593 0.39215686
 0.38709677 0.5        0.43478261 0.52       0.75757576 0.65079365
 0.59090909 0.57692308 0.7962963  0.1978022  0.30434783 0.19444444
 0.7816092  0.7745098  0.18421053 0.62962963 0.69421488 0.66
 0.71538462 0.65306122 0.73076923 0.59016393 0.39705882 0.4
 0.25842697 0.69924812 0.59433962 0.         0.32       0.5
 0.23809524 0.33333333 0.36363636 0.53125    0.08333333 0.7244898
 0.72727273 0.22222222 0.64       0.15454545 0.13333333 0.49074074
 0.11111111 0.24242424 0.56338028 0.38461538 0.51612903 0.32142857
 0.47368421 0.56       0.76666667 0.48387097 0.35294118 0.47058824
 0.40909091 0.82926829 0.11428571 0.14285714 0.43333333 0.21052632
 0.54285714 0.34482759 0.62162162 0.63492063 0.16666667 0.10714286
 0.67567568 0.7804878  0.55813953 0.62037037 0.73831776 0.25806452
 0.07407407 0.14285714 0.55555556 0.41176471 0.31632653 0.58333333
 0.234375   0.59183673 0.6097561  0.62857143 0.52808989 0.72463768
 0.49122807 0.34065934 0.60784314 0.05882353 0.66216216 0.42857143
 0.60377358 0.4        0.34545455 0.78723404 0.34146341 0.58333333
 0.85227273 0.58823529 0.36363636 0.475      0.30555556 0.48979592
 0.50943396 0.51851852 0.12413793 0.63461538 0.32       0.25925926
 0.26315789 0.26666667 0.66887417 0.43103448 0.07142857 0.38333333
 0.18644068 0.61842105 0.66       0.37209302 0.31818182 0.49152542
 0.57333333 0.41549296 0.58490566 0.11940299 0.546875   0.792
 0.67878788 0.27419355 0.28333333 0.83333333 0.51190476 0.25490196
 0.29508197 0.31428571 0.46534653 0.34375    0.42592593 0.35
 0.3        0.34482759 0.70588235 0.75757576 0.9223301  0.87735849
 0.76851852 0.72972973 0.37142857 0.6618705  0.68613139 0.68421053]
upper bound: 0.48195690400802077
-----Evaluation is finished------
Class Accuracy 46.97%
Overall Prec@1 50.72% Prec@5 79.04%
/home/hs/anaconda3/envs/dzl/lib/python3.7/site-packages/numpy/core/_asarray.py:136: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray
  return array(a, dtype, copy=False, order=order, subok=True)

#PAN Full

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
# python test_models.py something --VAP --batch_size=16 -j=4 --test_crops=1 --test_segments=8,8,8,8 --weights=pretrained/PAN_RGB_something_resnet50_shift8_blockres_avg_segment8_e50.pth.tar,pretrained/PAN_PA_something_resnet50_shift8_blockres_avg_segment8_e80.pth.tar,pretrained/PAN_PA_something_resnet50_shift8_blockres_avg_segment8_e80.pth.tar,pretrained/PAN_RGB_something_resnet50_shift8_blockres_avg_segment8_e50.pth.tar
$ bash scripts/test/sthv1/AAA.sh
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     RGB
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256

=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
video number:11522
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     PA
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256

=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
video number:11522
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     PA
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256

=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
video number:11522
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     RGB
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256

=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
video number:11522
video 0 done, total 0/11522, average 0.237 sec/video, moving Prec@1 50.000 Prec@5 87.500
video 320 done, total 320/11522, average 0.056 sec/video, moving Prec@1 51.488 Prec@5 83.333
video 640 done, total 640/11522, average 0.052 sec/video, moving Prec@1 52.287 Prec@5 82.165
video 960 done, total 960/11522, average 0.050 sec/video, moving Prec@1 51.332 Prec@5 80.840
video 1280 done, total 1280/11522, average 0.049 sec/video, moving Prec@1 52.160 Prec@5 81.096
video 1600 done, total 1600/11522, average 0.049 sec/video, moving Prec@1 51.795 Prec@5 80.446
video 1920 done, total 1920/11522, average 0.048 sec/video, moving Prec@1 50.878 Prec@5 80.062
video 2240 done, total 2240/11522, average 0.048 sec/video, moving Prec@1 50.798 Prec@5 79.920
video 2560 done, total 2560/11522, average 0.048 sec/video, moving Prec@1 50.970 Prec@5 79.775
video 2880 done, total 2880/11522, average 0.048 sec/video, moving Prec@1 50.760 Prec@5 79.731
video 3200 done, total 3200/11522, average 0.048 sec/video, moving Prec@1 50.529 Prec@5 79.695
video 3520 done, total 3520/11522, average 0.048 sec/video, moving Prec@1 50.509 Prec@5 79.638
video 3840 done, total 3840/11522, average 0.048 sec/video, moving Prec@1 50.467 Prec@5 79.279
video 4160 done, total 4160/11522, average 0.048 sec/video, moving Prec@1 50.359 Prec@5 79.215
video 4480 done, total 4480/11522, average 0.048 sec/video, moving Prec@1 50.311 Prec@5 79.248
video 4800 done, total 4800/11522, average 0.048 sec/video, moving Prec@1 50.104 Prec@5 78.987
video 5120 done, total 5120/11522, average 0.048 sec/video, moving Prec@1 49.883 Prec@5 78.777
video 5440 done, total 5440/11522, average 0.048 sec/video, moving Prec@1 50.202 Prec@5 78.886
video 5760 done, total 5760/11522, average 0.048 sec/video, moving Prec@1 50.173 Prec@5 78.722
video 6080 done, total 6080/11522, average 0.048 sec/video, moving Prec@1 50.262 Prec@5 78.888
video 6400 done, total 6400/11522, average 0.047 sec/video, moving Prec@1 50.327 Prec@5 79.068
video 6720 done, total 6720/11522, average 0.047 sec/video, moving Prec@1 50.238 Prec@5 79.053
video 7040 done, total 7040/11522, average 0.047 sec/video, moving Prec@1 50.241 Prec@5 78.954
video 7360 done, total 7360/11522, average 0.047 sec/video, moving Prec@1 50.176 Prec@5 78.823
video 7680 done, total 7680/11522, average 0.047 sec/video, moving Prec@1 50.299 Prec@5 78.976
video 8000 done, total 8000/11522, average 0.047 sec/video, moving Prec@1 50.274 Prec@5 79.067
video 8320 done, total 8320/11522, average 0.047 sec/video, moving Prec@1 50.288 Prec@5 79.115
video 8640 done, total 8640/11522, average 0.047 sec/video, moving Prec@1 50.139 Prec@5 79.136
video 8960 done, total 8960/11522, average 0.047 sec/video, moving Prec@1 50.212 Prec@5 79.144
video 9280 done, total 9280/11522, average 0.047 sec/video, moving Prec@1 50.237 Prec@5 79.131
video 9600 done, total 9600/11522, average 0.047 sec/video, moving Prec@1 50.374 Prec@5 79.181
video 9920 done, total 9920/11522, average 0.047 sec/video, moving Prec@1 50.312 Prec@5 79.177
video 10240 done, total 10240/11522, average 0.047 sec/video, moving Prec@1 50.429 Prec@5 79.329
video 10560 done, total 10560/11522, average 0.047 sec/video, moving Prec@1 50.454 Prec@5 79.283
video 10880 done, total 10880/11522, average 0.047 sec/video, moving Prec@1 50.569 Prec@5 79.323
video 11200 done, total 11200/11522, average 0.047 sec/video, moving Prec@1 50.606 Prec@5 79.271
video 11520 done, total 11520/11522, average 0.047 sec/video, moving Prec@1 50.529 Prec@5 79.231
[0.89552239 0.3125     0.26666667 0.65671642 0.36842105 0.48275862
 0.70895522 0.38235294 0.62135922 0.70873786 0.44       0.40277778
 0.30851064 0.35483871 0.77070064 0.4        0.24096386 0.2625
 0.45528455 0.34666667 0.352      0.60344828 0.46296296 0.45098039
 0.51612903 0.41666667 0.43478261 0.4        0.6969697  0.61904762
 0.59090909 0.53846154 0.87037037 0.25274725 0.23913043 0.13888889
 0.7816092  0.79411765 0.07894737 0.59259259 0.68595041 0.64
 0.67692308 0.67346939 0.75       0.57377049 0.47794118 0.36842105
 0.15730337 0.68421053 0.61320755 0.         0.36       0.45
 0.19047619 0.25       0.45454545 0.5        0.         0.74489796
 0.75757576 0.13888889 0.64       0.1        0.1        0.46296296
 0.18518519 0.21212121 0.43661972 0.34615385 0.5483871  0.21428571
 0.55263158 0.48       0.83333333 0.41935484 0.35294118 0.58823529
 0.27272727 0.82926829 0.11428571 0.21428571 0.33333333 0.15789474
 0.6        0.39655172 0.60810811 0.6984127  0.27777778 0.07142857
 0.59459459 0.75609756 0.58139535 0.55555556 0.62616822 0.22580645
 0.03703704 0.11904762 0.57894737 0.45588235 0.32653061 0.51851852
 0.203125   0.53061224 0.68292683 0.63809524 0.52808989 0.73913043
 0.49122807 0.32967033 0.58823529 0.         0.66216216 0.28571429
 0.60377358 0.43333333 0.27272727 0.73404255 0.29268293 0.54166667
 0.80681818 0.60784314 0.35454545 0.5        0.33333333 0.44897959
 0.47169811 0.57407407 0.14482759 0.63461538 0.22       0.22222222
 0.23684211 0.36666667 0.68874172 0.44827586 0.         0.31666667
 0.13559322 0.56578947 0.66       0.39534884 0.45454545 0.43220339
 0.6        0.47183099 0.63207547 0.17910448 0.546875   0.8
 0.66666667 0.29032258 0.38333333 0.7962963  0.55952381 0.11764706
 0.29508197 0.3        0.47524752 0.34375    0.42592593 0.325
 0.3        0.37931034 0.69607843 0.77272727 0.9223301  0.87735849
 0.74074074 0.75675676 0.45714286 0.62589928 0.69343066 0.69736842]
upper bound: 0.483756929950145
-----Evaluation is finished------
Class Accuracy 46.26%
Overall Prec@1 50.53% Prec@5 79.23%
/home/hs/anaconda3/envs/dzl/lib/python3.7/site-packages/numpy/core/_asarray.py:136: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray
  return array(a, dtype, copy=False, order=order, subok=True)

#PAN MyEn

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
# python test_models.py something --VAP --batch_size=10 -j=4 --test_crops=1 --test_segments=8,8,8,8 --weights=pretrained/PAN_RGB_something_resnet50_shift8_blockres_avg_segment8_e50.pth.tar,pretrained/PAN_PA_something_resnet50_shift8_blockres_avg_segment8_e80.pth.tar,pretrained/PAN_RPA_something_resnet50_shift8_blockres_avg_segment8_e80.pth.tar,pretrained/PAN_RGB_something_resnet50_shift8_blockres_avg_segment8_e50.pth.tar --full_res --twice_sample
$ bash scripts/test/sthv1/AAA.sh
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     RGB
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256

=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
=> Using twice sample for the dataset...
video number:11522
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     PA
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256

=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
=> Using twice sample for the dataset...
video number:11522
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     PA
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256

=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
=> Using twice sample for the dataset...
video number:11522
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     RGB
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256

=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
=> Using twice sample for the dataset...
video number:11522
video 0 done, total 0/11522, average 0.844 sec/video, moving Prec@1 60.000 Prec@5 80.000
video 200 done, total 200/11522, average 0.154 sec/video, moving Prec@1 58.571 Prec@5 83.810
video 400 done, total 400/11522, average 0.153 sec/video, moving Prec@1 56.829 Prec@5 83.902
video 600 done, total 600/11522, average 0.153 sec/video, moving Prec@1 57.541 Prec@5 83.279
video 800 done, total 800/11522, average 0.151 sec/video, moving Prec@1 54.815 Prec@5 81.481
video 1000 done, total 1000/11522, average 0.151 sec/video, moving Prec@1 54.950 Prec@5 80.990
video 1200 done, total 1200/11522, average 0.152 sec/video, moving Prec@1 55.207 Prec@5 81.901
video 1400 done, total 1400/11522, average 0.151 sec/video, moving Prec@1 55.390 Prec@5 81.986
video 1600 done, total 1600/11522, average 0.152 sec/video, moving Prec@1 55.404 Prec@5 81.801
video 1800 done, total 1800/11522, average 0.152 sec/video, moving Prec@1 54.530 Prec@5 81.271
video 2000 done, total 2000/11522, average 0.151 sec/video, moving Prec@1 54.677 Prec@5 81.542
video 2200 done, total 2200/11522, average 0.152 sec/video, moving Prec@1 54.525 Prec@5 81.629
video 2400 done, total 2400/11522, average 0.152 sec/video, moving Prec@1 54.149 Prec@5 81.203
video 2600 done, total 2600/11522, average 0.152 sec/video, moving Prec@1 54.100 Prec@5 81.188
video 2800 done, total 2800/11522, average 0.152 sec/video, moving Prec@1 53.772 Prec@5 81.174
video 3000 done, total 3000/11522, average 0.152 sec/video, moving Prec@1 53.654 Prec@5 81.462
video 3200 done, total 3200/11522, average 0.152 sec/video, moving Prec@1 53.271 Prec@5 81.277
video 3400 done, total 3400/11522, average 0.152 sec/video, moving Prec@1 53.343 Prec@5 81.525
video 3600 done, total 3600/11522, average 0.152 sec/video, moving Prec@1 53.269 Prec@5 81.191
video 3800 done, total 3800/11522, average 0.152 sec/video, moving Prec@1 53.255 Prec@5 81.050
video 4000 done, total 4000/11522, average 0.151 sec/video, moving Prec@1 53.117 Prec@5 80.848
video 4200 done, total 4200/11522, average 0.151 sec/video, moving Prec@1 53.064 Prec@5 80.879
video 4400 done, total 4400/11522, average 0.151 sec/video, moving Prec@1 52.857 Prec@5 80.816
video 4600 done, total 4600/11522, average 0.151 sec/video, moving Prec@1 52.907 Prec@5 80.781
video 4800 done, total 4800/11522, average 0.151 sec/video, moving Prec@1 52.786 Prec@5 80.707
video 5000 done, total 5000/11522, average 0.151 sec/video, moving Prec@1 52.455 Prec@5 80.399
video 5200 done, total 5200/11522, average 0.151 sec/video, moving Prec@1 52.534 Prec@5 80.557
video 5400 done, total 5400/11522, average 0.151 sec/video, moving Prec@1 52.699 Prec@5 80.573
video 5600 done, total 5600/11522, average 0.151 sec/video, moving Prec@1 52.852 Prec@5 80.624
video 5800 done, total 5800/11522, average 0.151 sec/video, moving Prec@1 52.616 Prec@5 80.448
video 6000 done, total 6000/11522, average 0.151 sec/video, moving Prec@1 52.679 Prec@5 80.532
video 6200 done, total 6200/11522, average 0.151 sec/video, moving Prec@1 52.625 Prec@5 80.499
video 6400 done, total 6400/11522, average 0.151 sec/video, moving Prec@1 52.746 Prec@5 80.718
video 6600 done, total 6600/11522, average 0.151 sec/video, moving Prec@1 52.617 Prec@5 80.756
video 6800 done, total 6800/11522, average 0.151 sec/video, moving Prec@1 52.555 Prec@5 80.675
video 7000 done, total 7000/11522, average 0.151 sec/video, moving Prec@1 52.511 Prec@5 80.685
video 7200 done, total 7200/11522, average 0.151 sec/video, moving Prec@1 52.413 Prec@5 80.596
video 7400 done, total 7400/11522, average 0.151 sec/video, moving Prec@1 52.470 Prec@5 80.553
video 7600 done, total 7600/11522, average 0.151 sec/video, moving Prec@1 52.549 Prec@5 80.631
video 7800 done, total 7800/11522, average 0.151 sec/video, moving Prec@1 52.510 Prec@5 80.602
video 8000 done, total 8000/11522, average 0.151 sec/video, moving Prec@1 52.447 Prec@5 80.587
video 8200 done, total 8200/11522, average 0.151 sec/video, moving Prec@1 52.460 Prec@5 80.597
video 8400 done, total 8400/11522, average 0.151 sec/video, moving Prec@1 52.556 Prec@5 80.618
video 8600 done, total 8600/11522, average 0.151 sec/video, moving Prec@1 52.427 Prec@5 80.604
video 8800 done, total 8800/11522, average 0.151 sec/video, moving Prec@1 52.486 Prec@5 80.681
video 9000 done, total 9000/11522, average 0.151 sec/video, moving Prec@1 52.431 Prec@5 80.655
video 9200 done, total 9200/11522, average 0.151 sec/video, moving Prec@1 52.476 Prec@5 80.662
video 9400 done, total 9400/11522, average 0.151 sec/video, moving Prec@1 52.508 Prec@5 80.638
video 9600 done, total 9600/11522, average 0.151 sec/video, moving Prec@1 52.591 Prec@5 80.666
video 9800 done, total 9800/11522, average 0.151 sec/video, moving Prec@1 52.589 Prec@5 80.693
video 10000 done, total 10000/11522, average 0.151 sec/video, moving Prec@1 52.667 Prec@5 80.689
video 10200 done, total 10200/11522, average 0.151 sec/video, moving Prec@1 52.752 Prec@5 80.695
video 10400 done, total 10400/11522, average 0.151 sec/video, moving Prec@1 52.786 Prec@5 80.663
video 10600 done, total 10600/11522, average 0.151 sec/video, moving Prec@1 52.762 Prec@5 80.669
video 10800 done, total 10800/11522, average 0.151 sec/video, moving Prec@1 52.840 Prec@5 80.712
video 11000 done, total 11000/11522, average 0.151 sec/video, moving Prec@1 52.861 Prec@5 80.708
video 11200 done, total 11200/11522, average 0.151 sec/video, moving Prec@1 52.899 Prec@5 80.740
video 11400 done, total 11400/11522, average 0.151 sec/video, moving Prec@1 52.875 Prec@5 80.710
[0.91044776 0.4        0.28571429 0.68656716 0.47368421 0.46551724
 0.69402985 0.44117647 0.74757282 0.7184466  0.45333333 0.5
 0.37234043 0.41935484 0.77070064 0.5        0.21084337 0.2875
 0.4796748  0.38666667 0.368      0.62068966 0.5        0.50980392
 0.51612903 0.5        0.43478261 0.44       0.72727273 0.6984127
 0.72727273 0.5        0.83333333 0.24175824 0.26086957 0.22222222
 0.81609195 0.81372549 0.05263158 0.66666667 0.71900826 0.62
 0.73846154 0.67346939 0.69230769 0.62295082 0.48529412 0.41052632
 0.24719101 0.7518797  0.62264151 0.         0.4        0.5
 0.19047619 0.16666667 0.42424242 0.4375     0.         0.75510204
 0.6969697  0.22222222 0.64       0.14545455 0.2        0.5
 0.12962963 0.25757576 0.52112676 0.38461538 0.58064516 0.32142857
 0.5        0.48       0.83333333 0.5483871  0.23529412 0.58823529
 0.40909091 0.82926829 0.08571429 0.14285714 0.4        0.26315789
 0.57142857 0.43103448 0.66216216 0.79365079 0.11111111 0.125
 0.62162162 0.73170732 0.58139535 0.68518519 0.71962617 0.32258065
 0.03703704 0.16666667 0.56725146 0.42647059 0.39795918 0.5462963
 0.21875    0.57142857 0.68292683 0.65714286 0.52808989 0.73913043
 0.42105263 0.34065934 0.64705882 0.11764706 0.67567568 0.33333333
 0.66037736 0.5        0.32727273 0.78723404 0.34146341 0.58333333
 0.82954545 0.58823529 0.39090909 0.475      0.30555556 0.53061224
 0.52830189 0.61111111 0.13793103 0.66346154 0.28       0.22222222
 0.28947368 0.33333333 0.72847682 0.39655172 0.         0.26666667
 0.13559322 0.60526316 0.67       0.44186047 0.40909091 0.45762712
 0.6        0.47183099 0.62264151 0.19402985 0.59375    0.76
 0.68484848 0.27419355 0.4        0.88888889 0.57142857 0.23529412
 0.26229508 0.35714286 0.5049505  0.34375    0.44444444 0.375
 0.32       0.37931034 0.66666667 0.75757576 0.9223301  0.88679245
 0.76851852 0.67567568 0.44761905 0.66906475 0.7080292  0.73684211]
upper bound: 0.5027293241199311
-----Evaluation is finished------
Class Accuracy 48.55%
Overall Prec@1 52.86% Prec@5 80.72%
/home/hs/anaconda3/envs/dzl/lib/python3.7/site-packages/numpy/core/_asarray.py:136: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray
  return array(a, dtype, copy=False, order=order, subok=True)

#PAN RGB&PA&RPA

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
# python test_models.py something \
# --VAP --batch_size=16 -j=4 --test_crops=1 --test_segments=8,8,8 \
# --weights=pretrained/PAN_RPA_something_resnet50_shift8_blockres_avg_segment8_e80.pth.tar,pretrained/PAN_RGB_something_resnet50_shift8_blockres_avg_segment8_e50.pth.tar,pretrained/PAN_PA_something_resnet50_shift8_blockres_avg_segment8_e80.pth.tar \
$ bash scripts/test/sthv1/Three.sh 
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     PA
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256
            
=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
video number:11522
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     RGB
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256
            
=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
video number:11522
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     PA
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256
            
=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
video number:11522
video 0 done, total 0/11522, average 0.267 sec/video, moving Prec@1 56.250 Prec@5 87.500
video 320 done, total 320/11522, average 0.052 sec/video, moving Prec@1 54.167 Prec@5 83.631
video 640 done, total 640/11522, average 0.048 sec/video, moving Prec@1 54.573 Prec@5 82.622
video 960 done, total 960/11522, average 0.047 sec/video, moving Prec@1 53.484 Prec@5 81.557
video 1280 done, total 1280/11522, average 0.046 sec/video, moving Prec@1 54.321 Prec@5 81.559
video 1600 done, total 1600/11522, average 0.046 sec/video, moving Prec@1 54.022 Prec@5 80.879
video 1920 done, total 1920/11522, average 0.045 sec/video, moving Prec@1 53.099 Prec@5 80.527
video 2240 done, total 2240/11522, average 0.045 sec/video, moving Prec@1 52.793 Prec@5 80.496
video 2560 done, total 2560/11522, average 0.045 sec/video, moving Prec@1 52.601 Prec@5 80.551
video 2880 done, total 2880/11522, average 0.045 sec/video, moving Prec@1 52.279 Prec@5 80.318
video 3200 done, total 3200/11522, average 0.044 sec/video, moving Prec@1 51.959 Prec@5 80.504
video 3520 done, total 3520/11522, average 0.044 sec/video, moving Prec@1 51.810 Prec@5 80.317
video 3840 done, total 3840/11522, average 0.044 sec/video, moving Prec@1 51.608 Prec@5 80.005
video 4160 done, total 4160/11522, average 0.044 sec/video, moving Prec@1 51.389 Prec@5 79.861
video 4480 done, total 4480/11522, average 0.044 sec/video, moving Prec@1 51.268 Prec@5 79.738
video 4800 done, total 4800/11522, average 0.044 sec/video, moving Prec@1 50.997 Prec@5 79.527
video 5120 done, total 5120/11522, average 0.044 sec/video, moving Prec@1 50.759 Prec@5 79.264
video 5440 done, total 5440/11522, average 0.044 sec/video, moving Prec@1 50.898 Prec@5 79.399
video 5760 done, total 5760/11522, average 0.044 sec/video, moving Prec@1 50.779 Prec@5 79.224
video 6080 done, total 6080/11522, average 0.044 sec/video, moving Prec@1 50.820 Prec@5 79.331
video 6400 done, total 6400/11522, average 0.043 sec/video, moving Prec@1 50.935 Prec@5 79.458
video 6720 done, total 6720/11522, average 0.043 sec/video, moving Prec@1 50.861 Prec@5 79.454
video 7040 done, total 7040/11522, average 0.043 sec/video, moving Prec@1 50.893 Prec@5 79.408
video 7360 done, total 7360/11522, average 0.043 sec/video, moving Prec@1 50.813 Prec@5 79.284
video 7680 done, total 7680/11522, average 0.043 sec/video, moving Prec@1 50.975 Prec@5 79.431
video 8000 done, total 8000/11522, average 0.043 sec/video, moving Prec@1 50.948 Prec@5 79.479
video 8320 done, total 8320/11522, average 0.043 sec/video, moving Prec@1 50.936 Prec@5 79.559
video 8640 done, total 8640/11522, average 0.043 sec/video, moving Prec@1 50.739 Prec@5 79.552
video 8960 done, total 8960/11522, average 0.043 sec/video, moving Prec@1 50.780 Prec@5 79.612
video 9280 done, total 9280/11522, average 0.043 sec/video, moving Prec@1 50.818 Prec@5 79.561
video 9600 done, total 9600/11522, average 0.043 sec/video, moving Prec@1 50.915 Prec@5 79.555
video 9920 done, total 9920/11522, average 0.043 sec/video, moving Prec@1 50.946 Prec@5 79.569
video 10240 done, total 10240/11522, average 0.043 sec/video, moving Prec@1 50.985 Prec@5 79.670
video 10560 done, total 10560/11522, average 0.043 sec/video, moving Prec@1 51.059 Prec@5 79.643
video 10880 done, total 10880/11522, average 0.043 sec/video, moving Prec@1 51.120 Prec@5 79.690
video 11200 done, total 11200/11522, average 0.043 sec/video, moving Prec@1 51.221 Prec@5 79.654
video 11520 done, total 11520/11522, average 0.043 sec/video, moving Prec@1 51.120 Prec@5 79.622
[0.91044776 0.3625     0.3047619  0.64179104 0.39473684 0.46551724
 0.67164179 0.41176471 0.6407767  0.67961165 0.41333333 0.47222222
 0.31914894 0.4516129  0.70700637 0.44       0.25903614 0.275
 0.44715447 0.4        0.352      0.5862069  0.5        0.43137255
 0.5483871  0.41666667 0.43478261 0.52       0.72727273 0.65079365
 0.59090909 0.53846154 0.81481481 0.21978022 0.23913043 0.11111111
 0.83908046 0.78431373 0.18421053 0.62962963 0.69421488 0.7
 0.67692308 0.69387755 0.71153846 0.62295082 0.44852941 0.37894737
 0.21348315 0.69172932 0.59433962 0.         0.4        0.45
 0.23809524 0.29166667 0.42424242 0.5625     0.         0.7244898
 0.78787879 0.19444444 0.62       0.18181818 0.1        0.46296296
 0.18518519 0.22727273 0.47887324 0.30769231 0.5483871  0.25
 0.47368421 0.48       0.8        0.41935484 0.35294118 0.52941176
 0.40909091 0.80487805 0.11428571 0.23809524 0.43333333 0.15789474
 0.57142857 0.39655172 0.63513514 0.66666667 0.16666667 0.10714286
 0.59459459 0.7804878  0.48837209 0.61111111 0.70093458 0.25806452
 0.07407407 0.16666667 0.58479532 0.48529412 0.33673469 0.5462963
 0.1875     0.63265306 0.65853659 0.61904762 0.46067416 0.73913043
 0.49122807 0.34065934 0.58823529 0.         0.66216216 0.33333333
 0.56603774 0.36666667 0.34545455 0.79787234 0.29268293 0.5625
 0.79545455 0.60784314 0.36363636 0.5        0.27777778 0.42857143
 0.49056604 0.62962963 0.13793103 0.63461538 0.26       0.22222222
 0.23684211 0.4        0.65562914 0.4137931  0.07142857 0.3
 0.11864407 0.63157895 0.67       0.37209302 0.45454545 0.48305085
 0.57333333 0.47887324 0.63207547 0.13432836 0.53125    0.8
 0.68484848 0.29032258 0.33333333 0.83333333 0.52380952 0.23529412
 0.26229508 0.34285714 0.46534653 0.28125    0.48148148 0.325
 0.34       0.37931034 0.7254902  0.75757576 0.9223301  0.86792453
 0.74074074 0.7027027  0.38095238 0.64028777 0.70072993 0.71052632]
upper bound: 0.48707027638701955
-----Evaluation is finished------
Class Accuracy 47.08%
Overall Prec@1 51.12% Prec@5 79.62%
/home/hs/anaconda3/envs/dzl/lib/python3.7/site-packages/numpy/core/_asarray.py:136: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray
  return array(a, dtype, copy=False, order=order, subok=True)

#20201009 RGB&PA&RPA&RGB

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
# python test_models.py something --VAP --batch_size=16 -j=4 --test_crops=1 --test_segments=8,8,8,8 --weights=pretrained/PAN_RGB_something_resnet50_shift8_blockres_avg_segment8_e50.pth.tar,pretrained/PAN_PA_something_resnet50_shift8_blockres_avg_segment8_e80.pth.tar,pretrained/PAN_RPA_something_resnet50_shift8_blockres_avg_segment8_e80.pth.tar,pretrained/PAN_RGB_something_resnet50_shift8_blockres_avg_segment8_e50.pth.tar
$ bash scripts/test/sthv1/AAA.sh 
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     RGB
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256
            
=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
video number:11522
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     PA
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256
            
=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
video number:11522
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     PA
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256
            
=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
video number:11522
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     RGB
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256
            
=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
video number:11522
video 0 done, total 0/11522, average 0.221 sec/video, moving Prec@1 56.250 Prec@5 87.500
video 320 done, total 320/11522, average 0.055 sec/video, moving Prec@1 55.060 Prec@5 83.036
video 640 done, total 640/11522, average 0.053 sec/video, moving Prec@1 55.335 Prec@5 82.012
video 960 done, total 960/11522, average 0.052 sec/video, moving Prec@1 54.303 Prec@5 80.840
video 1280 done, total 1280/11522, average 0.051 sec/video, moving Prec@1 54.552 Prec@5 81.173
video 1600 done, total 1600/11522, average 0.050 sec/video, moving Prec@1 54.455 Prec@5 80.631
video 1920 done, total 1920/11522, average 0.050 sec/video, moving Prec@1 53.409 Prec@5 80.217
video 2240 done, total 2240/11522, average 0.050 sec/video, moving Prec@1 53.236 Prec@5 80.319
video 2560 done, total 2560/11522, average 0.050 sec/video, moving Prec@1 53.106 Prec@5 80.435
video 2880 done, total 2880/11522, average 0.050 sec/video, moving Prec@1 52.624 Prec@5 80.249
video 3200 done, total 3200/11522, average 0.049 sec/video, moving Prec@1 52.239 Prec@5 80.224
video 3520 done, total 3520/11522, average 0.049 sec/video, moving Prec@1 52.206 Prec@5 80.119
video 3840 done, total 3840/11522, average 0.049 sec/video, moving Prec@1 52.127 Prec@5 79.798
video 4160 done, total 4160/11522, average 0.049 sec/video, moving Prec@1 51.796 Prec@5 79.765
video 4480 done, total 4480/11522, average 0.049 sec/video, moving Prec@1 51.646 Prec@5 79.760
video 4800 done, total 4800/11522, average 0.049 sec/video, moving Prec@1 51.474 Prec@5 79.360
video 5120 done, total 5120/11522, average 0.049 sec/video, moving Prec@1 51.188 Prec@5 79.186
video 5440 done, total 5440/11522, average 0.049 sec/video, moving Prec@1 51.338 Prec@5 79.362
video 5760 done, total 5760/11522, average 0.049 sec/video, moving Prec@1 51.229 Prec@5 79.190
video 6080 done, total 6080/11522, average 0.049 sec/video, moving Prec@1 51.280 Prec@5 79.331
video 6400 done, total 6400/11522, average 0.049 sec/video, moving Prec@1 51.372 Prec@5 79.520
video 6720 done, total 6720/11522, average 0.049 sec/video, moving Prec@1 51.292 Prec@5 79.528
video 7040 done, total 7040/11522, average 0.049 sec/video, moving Prec@1 51.361 Prec@5 79.478
video 7360 done, total 7360/11522, average 0.049 sec/video, moving Prec@1 51.315 Prec@5 79.366
video 7680 done, total 7680/11522, average 0.049 sec/video, moving Prec@1 51.494 Prec@5 79.548
video 8000 done, total 8000/11522, average 0.049 sec/video, moving Prec@1 51.422 Prec@5 79.591
video 8320 done, total 8320/11522, average 0.049 sec/video, moving Prec@1 51.476 Prec@5 79.667
video 8640 done, total 8640/11522, average 0.049 sec/video, moving Prec@1 51.340 Prec@5 79.656
video 8960 done, total 8960/11522, average 0.049 sec/video, moving Prec@1 51.326 Prec@5 79.690
video 9280 done, total 9280/11522, average 0.049 sec/video, moving Prec@1 51.366 Prec@5 79.669
video 9600 done, total 9600/11522, average 0.049 sec/video, moving Prec@1 51.466 Prec@5 79.732
video 9920 done, total 9920/11522, average 0.049 sec/video, moving Prec@1 51.449 Prec@5 79.740
video 10240 done, total 10240/11522, average 0.049 sec/video, moving Prec@1 51.511 Prec@5 79.846
video 10560 done, total 10560/11522, average 0.049 sec/video, moving Prec@1 51.560 Prec@5 79.870
video 10880 done, total 10880/11522, average 0.049 sec/video, moving Prec@1 51.597 Prec@5 79.956
video 11200 done, total 11200/11522, average 0.049 sec/video, moving Prec@1 51.649 Prec@5 79.939
video 11520 done, total 11520/11522, average 0.049 sec/video, moving Prec@1 51.536 Prec@5 79.865
[0.89552239 0.325      0.28571429 0.64179104 0.44736842 0.46551724
 0.70149254 0.41176471 0.63106796 0.70873786 0.41333333 0.48611111
 0.30851064 0.41935484 0.75159236 0.46       0.23493976 0.275
 0.42276423 0.34666667 0.344      0.62068966 0.5        0.43137255
 0.51612903 0.41666667 0.43478261 0.48       0.72727273 0.61904762
 0.63636364 0.57692308 0.83333333 0.25274725 0.32608696 0.16666667
 0.82758621 0.79411765 0.10526316 0.59259259 0.68595041 0.66
 0.70769231 0.67346939 0.69230769 0.57377049 0.46323529 0.41052632
 0.17977528 0.69924812 0.6509434  0.         0.36       0.4
 0.23809524 0.25       0.42424242 0.5625     0.         0.75510204
 0.81818182 0.19444444 0.66       0.18181818 0.1        0.49074074
 0.14814815 0.22727273 0.45070423 0.34615385 0.5483871  0.21428571
 0.5        0.52       0.81666667 0.51612903 0.35294118 0.52941176
 0.36363636 0.80487805 0.08571429 0.21428571 0.4        0.15789474
 0.6        0.37931034 0.59459459 0.65079365 0.27777778 0.07142857
 0.59459459 0.75609756 0.55813953 0.58333333 0.68224299 0.25806452
 0.03703704 0.11904762 0.56140351 0.5        0.30612245 0.5462963
 0.203125   0.6122449  0.65853659 0.63809524 0.53932584 0.73913043
 0.47368421 0.32967033 0.60784314 0.05882353 0.67567568 0.28571429
 0.62264151 0.43333333 0.38181818 0.78723404 0.31707317 0.58333333
 0.81818182 0.60784314 0.36363636 0.5        0.30555556 0.46938776
 0.45283019 0.62962963 0.15862069 0.625      0.28       0.25925926
 0.28947368 0.3        0.67549669 0.46551724 0.07142857 0.36666667
 0.15254237 0.56578947 0.67       0.39534884 0.38636364 0.46610169
 0.58666667 0.45774648 0.63207547 0.1641791  0.5625     0.824
 0.68484848 0.29032258 0.36666667 0.83333333 0.54761905 0.15686275
 0.29508197 0.31428571 0.48514851 0.375      0.5        0.425
 0.32       0.4137931  0.71568627 0.77272727 0.9223301  0.87735849
 0.75925926 0.7027027  0.45714286 0.62589928 0.69343066 0.71052632]
upper bound: 0.4923224364393845
-----Evaluation is finished------
Class Accuracy 47.49%
Overall Prec@1 51.54% Prec@5 79.86%
/home/hs/anaconda3/envs/dzl/lib/python3.7/site-packages/numpy/core/_asarray.py:136: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray
  return array(a, dtype, copy=False, order=order, subok=True)

#20201009 RGB&PA&RPA&RGB&Lite&Lite

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
# python test_models.py something --VAP --batch_size=8 -j=4 --test_crops=1 --test_segments=8,8,8,8,8,8 --weights=pretrained/PAN_RGB_something_resnet50_shift8_blockres_avg_segment8_e50.pth.tar,pretrained/PAN_PA_something_resnet50_shift8_blockres_avg_segment8_e80.pth.tar,pretrained/PAN_RPA_something_resnet50_shift8_blockres_avg_segment8_e80.pth.tar,pretrained/PAN_RGB_something_resnet50_shift8_blockres_avg_segment8_e50.pth.tar,pretrained/PAN_Lite_something_resnet50_shift8_blockres_avg_segment8_e80.pth.tar,pretrained/PAN_Lite_something_resnet50_shift8_blockres_avg_segment8_e80.pth.tar --full_res --twice_sample
$ bash scripts/test/sthv1/6.sh 
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     RGB
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256
            
=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
=> Using twice sample for the dataset...
video number:11522
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     PA
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256
            
=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
=> Using twice sample for the dataset...
video number:11522
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     PA
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256
            
=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
=> Using twice sample for the dataset...
video number:11522
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     RGB
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256
            
=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
=> Using twice sample for the dataset...
video number:11522
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     Lite
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256
            
=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
=> Converting the ImageNet model to a PAN_Lite init model
=> Done. PAN_lite model ready...
=> Using twice sample for the dataset...
video number:11522
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     Lite
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256
            
=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
=> Converting the ImageNet model to a PAN_Lite init model
=> Done. PAN_lite model ready...
=> Using twice sample for the dataset...
video number:11522
video 0 done, total 0/11522, average 1.315 sec/video, moving Prec@1 50.000 Prec@5 75.000
video 160 done, total 160/11522, average 0.300 sec/video, moving Prec@1 57.738 Prec@5 86.310
video 320 done, total 320/11522, average 0.314 sec/video, moving Prec@1 55.793 Prec@5 85.366
video 480 done, total 480/11522, average 0.318 sec/video, moving Prec@1 57.582 Prec@5 85.656
video 640 done, total 640/11522, average 0.318 sec/video, moving Prec@1 56.636 Prec@5 83.796
video 800 done, total 800/11522, average 0.322 sec/video, moving Prec@1 55.569 Prec@5 82.673
video 960 done, total 960/11522, average 0.322 sec/video, moving Prec@1 55.062 Prec@5 82.541
video 1120 done, total 1120/11522, average 0.321 sec/video, moving Prec@1 56.206 Prec@5 83.067
video 1280 done, total 1280/11522, average 0.321 sec/video, moving Prec@1 56.522 Prec@5 83.307
video 1440 done, total 1440/11522, average 0.320 sec/video, moving Prec@1 56.354 Prec@5 82.942
video 1600 done, total 1600/11522, average 0.320 sec/video, moving Prec@1 56.468 Prec@5 83.022
video 1760 done, total 1760/11522, average 0.320 sec/video, moving Prec@1 55.826 Prec@5 82.579
video 1920 done, total 1920/11522, average 0.320 sec/video, moving Prec@1 55.705 Prec@5 82.573
video 2080 done, total 2080/11522, average 0.321 sec/video, moving Prec@1 55.603 Prec@5 82.375
video 2240 done, total 2240/11522, average 0.321 sec/video, moving Prec@1 55.783 Prec@5 82.429
video 2400 done, total 2400/11522, average 0.321 sec/video, moving Prec@1 55.399 Prec@5 82.018
video 2560 done, total 2560/11522, average 0.321 sec/video, moving Prec@1 55.335 Prec@5 82.009
video 2720 done, total 2720/11522, average 0.321 sec/video, moving Prec@1 55.389 Prec@5 82.111
video 2880 done, total 2880/11522, average 0.320 sec/video, moving Prec@1 55.090 Prec@5 82.168
video 3040 done, total 3040/11522, average 0.320 sec/video, moving Prec@1 54.921 Prec@5 82.283
video 3200 done, total 3200/11522, average 0.321 sec/video, moving Prec@1 54.426 Prec@5 82.170
video 3360 done, total 3360/11522, average 0.320 sec/video, moving Prec@1 54.365 Prec@5 82.452
video 3520 done, total 3520/11522, average 0.321 sec/video, moving Prec@1 54.223 Prec@5 82.115
video 3680 done, total 3680/11522, average 0.321 sec/video, moving Prec@1 54.094 Prec@5 81.860
video 3840 done, total 3840/11522, average 0.321 sec/video, moving Prec@1 54.314 Prec@5 81.861
video 4000 done, total 4000/11522, average 0.321 sec/video, moving Prec@1 54.167 Prec@5 81.737
video 4160 done, total 4160/11522, average 0.321 sec/video, moving Prec@1 54.079 Prec@5 81.694
video 4320 done, total 4320/11522, average 0.321 sec/video, moving Prec@1 53.997 Prec@5 81.608
video 4480 done, total 4480/11522, average 0.321 sec/video, moving Prec@1 53.922 Prec@5 81.640
video 4640 done, total 4640/11522, average 0.321 sec/video, moving Prec@1 53.894 Prec@5 81.497
video 4800 done, total 4800/11522, average 0.321 sec/video, moving Prec@1 53.765 Prec@5 81.448
video 4960 done, total 4960/11522, average 0.321 sec/video, moving Prec@1 53.462 Prec@5 81.200
video 5120 done, total 5120/11522, average 0.321 sec/video, moving Prec@1 53.432 Prec@5 81.162
video 5280 done, total 5280/11522, average 0.322 sec/video, moving Prec@1 53.423 Prec@5 81.127
video 5440 done, total 5440/11522, average 0.321 sec/video, moving Prec@1 53.616 Prec@5 81.351
video 5600 done, total 5600/11522, average 0.321 sec/video, moving Prec@1 53.709 Prec@5 81.384
video 5760 done, total 5760/11522, average 0.321 sec/video, moving Prec@1 53.589 Prec@5 81.224
video 5920 done, total 5920/11522, average 0.321 sec/video, moving Prec@1 53.492 Prec@5 81.275
video 6080 done, total 6080/11522, average 0.322 sec/video, moving Prec@1 53.400 Prec@5 81.340
video 6240 done, total 6240/11522, average 0.322 sec/video, moving Prec@1 53.457 Prec@5 81.354
video 6400 done, total 6400/11522, average 0.322 sec/video, moving Prec@1 53.480 Prec@5 81.445
video 6560 done, total 6560/11522, average 0.321 sec/video, moving Prec@1 53.426 Prec@5 81.532
video 6720 done, total 6720/11522, average 0.321 sec/video, moving Prec@1 53.404 Prec@5 81.495
video 6880 done, total 6880/11522, average 0.321 sec/video, moving Prec@1 53.296 Prec@5 81.402
video 7040 done, total 7040/11522, average 0.321 sec/video, moving Prec@1 53.348 Prec@5 81.413
video 7200 done, total 7200/11522, average 0.321 sec/video, moving Prec@1 53.344 Prec@5 81.396
video 7360 done, total 7360/11522, average 0.321 sec/video, moving Prec@1 53.312 Prec@5 81.325
video 7520 done, total 7520/11522, average 0.321 sec/video, moving Prec@1 53.374 Prec@5 81.389
video 7680 done, total 7680/11522, average 0.322 sec/video, moving Prec@1 53.356 Prec@5 81.413
video 7840 done, total 7840/11522, average 0.322 sec/video, moving Prec@1 53.313 Prec@5 81.397
video 8000 done, total 8000/11522, average 0.322 sec/video, moving Prec@1 53.247 Prec@5 81.381
video 8160 done, total 8160/11522, average 0.322 sec/video, moving Prec@1 53.293 Prec@5 81.428
video 8320 done, total 8320/11522, average 0.321 sec/video, moving Prec@1 53.338 Prec@5 81.460
video 8480 done, total 8480/11522, average 0.322 sec/video, moving Prec@1 53.228 Prec@5 81.374
video 8640 done, total 8640/11522, average 0.322 sec/video, moving Prec@1 53.076 Prec@5 81.406
video 8800 done, total 8800/11522, average 0.322 sec/video, moving Prec@1 53.134 Prec@5 81.460
video 8960 done, total 8960/11522, average 0.322 sec/video, moving Prec@1 53.066 Prec@5 81.479
video 9120 done, total 9120/11522, average 0.321 sec/video, moving Prec@1 53.024 Prec@5 81.453
video 9280 done, total 9280/11522, average 0.321 sec/video, moving Prec@1 53.025 Prec@5 81.449
video 9440 done, total 9440/11522, average 0.322 sec/video, moving Prec@1 53.059 Prec@5 81.425
video 9600 done, total 9600/11522, average 0.321 sec/video, moving Prec@1 53.143 Prec@5 81.495
video 9760 done, total 9760/11522, average 0.321 sec/video, moving Prec@1 53.184 Prec@5 81.511
video 9920 done, total 9920/11522, average 0.322 sec/video, moving Prec@1 53.112 Prec@5 81.467
video 10080 done, total 10080/11522, average 0.322 sec/video, moving Prec@1 53.202 Prec@5 81.572
video 10240 done, total 10240/11522, average 0.322 sec/video, moving Prec@1 53.230 Prec@5 81.616
video 10400 done, total 10400/11522, average 0.322 sec/video, moving Prec@1 53.324 Prec@5 81.572
video 10560 done, total 10560/11522, average 0.322 sec/video, moving Prec@1 53.340 Prec@5 81.558
video 10720 done, total 10720/11522, average 0.322 sec/video, moving Prec@1 53.384 Prec@5 81.572
video 10880 done, total 10880/11522, average 0.322 sec/video, moving Prec@1 53.398 Prec@5 81.585
video 11040 done, total 11040/11522, average 0.322 sec/video, moving Prec@1 53.385 Prec@5 81.598
video 11200 done, total 11200/11522, average 0.322 sec/video, moving Prec@1 53.462 Prec@5 81.620
video 11360 done, total 11360/11522, average 0.322 sec/video, moving Prec@1 53.439 Prec@5 81.606
video 11520 done, total 11520/11522, average 0.321 sec/video, moving Prec@1 53.524 Prec@5 81.592
[0.89552239 0.4375     0.3047619  0.67164179 0.47368421 0.49137931
 0.69402985 0.5        0.73786408 0.73786408 0.52       0.52777778
 0.41489362 0.41935484 0.8089172  0.48       0.24698795 0.35
 0.49593496 0.37333333 0.416      0.63793103 0.44444444 0.41176471
 0.51612903 0.41666667 0.52173913 0.4        0.6969697  0.74603175
 0.63636364 0.53846154 0.87037037 0.25274725 0.26086957 0.16666667
 0.8045977  0.85294118 0.07894737 0.62962963 0.7107438  0.66
 0.72307692 0.69387755 0.71153846 0.67213115 0.47058824 0.43157895
 0.29213483 0.72932331 0.61320755 0.         0.24       0.55
 0.23809524 0.20833333 0.51515152 0.5        0.         0.7755102
 0.78787879 0.25       0.62       0.12727273 0.16666667 0.55555556
 0.11111111 0.24242424 0.43661972 0.42307692 0.51612903 0.25
 0.47368421 0.56       0.83333333 0.58064516 0.29411765 0.47058824
 0.40909091 0.82926829 0.11428571 0.11904762 0.43333333 0.15789474
 0.62857143 0.39655172 0.64864865 0.77777778 0.11111111 0.08928571
 0.59459459 0.75609756 0.55813953 0.61111111 0.75700935 0.25806452
 0.03703704 0.16666667 0.60818713 0.51470588 0.3877551  0.58333333
 0.234375   0.53061224 0.70731707 0.6952381  0.62921348 0.75362319
 0.43859649 0.36263736 0.62745098 0.11764706 0.64864865 0.38095238
 0.66981132 0.46666667 0.32727273 0.78723404 0.31707317 0.60416667
 0.81818182 0.60784314 0.41818182 0.45       0.27777778 0.51020408
 0.45283019 0.62962963 0.17241379 0.68269231 0.3        0.18518519
 0.39473684 0.26666667 0.70198675 0.46551724 0.14285714 0.2
 0.15254237 0.65789474 0.66       0.3255814  0.52272727 0.48305085
 0.54666667 0.50704225 0.64150943 0.20895522 0.53125    0.736
 0.67272727 0.29032258 0.35       0.88888889 0.5952381  0.19607843
 0.24590164 0.35714286 0.53465347 0.34375    0.48148148 0.375
 0.36       0.4137931  0.69607843 0.81818182 0.9223301  0.91509434
 0.74074074 0.7027027  0.47619048 0.61151079 0.7080292  0.71052632]
upper bound: 0.5069670549400674
-----Evaluation is finished------
Class Accuracy 48.98%
Overall Prec@1 53.52% Prec@5 81.59%
/home/hs/anaconda3/envs/dzl/lib/python3.7/site-packages/numpy/core/_asarray.py:136: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray
  return array(a, dtype, copy=False, order=order, subok=True)

#20201009 RGB&PA&Lite（论文结果）

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
# python test_models.py something --VAP --batch_size=16 -j=4 --test_crops=1 --test_segments=8,8,8 --weights=o/PAN_Lite_something_resnet50_shift8_blockres_avg_segment8_e80.pth.tar,o/PAN_RGB_something_resnet50_shift8_blockres_avg_segment8_e50.pth.tar,o/PAN_PA_something_resnet50_shift8_blockres_avg_segment8_e80.pth.tar --full_res --twice_sample
$ bash scripts/test/sthv1/En.sh 
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     Lite
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256
            
=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
=> Converting the ImageNet model to a PAN_Lite init model
=> Done. PAN_lite model ready...
=> Using twice sample for the dataset...
video number:11522
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     RGB
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256
            
=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
=> Using twice sample for the dataset...
video number:11522
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     PA
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256
            
=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
=> Using twice sample for the dataset...
video number:11522
video 0 done, total 0/11522, average 1.100 sec/video, moving Prec@1 50.000 Prec@5 75.000
video 320 done, total 320/11522, average 0.158 sec/video, moving Prec@1 58.333 Prec@5 84.226
video 640 done, total 640/11522, average 0.152 sec/video, moving Prec@1 57.317 Prec@5 83.384
video 960 done, total 960/11522, average 0.154 sec/video, moving Prec@1 55.738 Prec@5 81.865
video 1280 done, total 1280/11522, average 0.152 sec/video, moving Prec@1 56.327 Prec@5 82.639
video 1600 done, total 1600/11522, average 0.151 sec/video, moving Prec@1 56.436 Prec@5 82.364
video 1920 done, total 1920/11522, average 0.150 sec/video, moving Prec@1 55.527 Prec@5 81.870
video 2240 done, total 2240/11522, average 0.150 sec/video, moving Prec@1 55.230 Prec@5 81.826
video 2560 done, total 2560/11522, average 0.149 sec/video, moving Prec@1 54.775 Prec@5 81.677
video 2880 done, total 2880/11522, average 0.149 sec/video, moving Prec@1 54.489 Prec@5 81.802
video 3200 done, total 3200/11522, average 0.149 sec/video, moving Prec@1 54.073 Prec@5 81.872
video 3520 done, total 3520/11522, average 0.149 sec/video, moving Prec@1 53.761 Prec@5 81.900
video 3840 done, total 3840/11522, average 0.149 sec/video, moving Prec@1 53.864 Prec@5 81.483
video 4160 done, total 4160/11522, average 0.149 sec/video, moving Prec@1 53.736 Prec@5 81.418
video 4480 done, total 4480/11522, average 0.149 sec/video, moving Prec@1 53.536 Prec@5 81.539
video 4800 done, total 4800/11522, average 0.148 sec/video, moving Prec@1 53.447 Prec@5 81.292
video 5120 done, total 5120/11522, average 0.149 sec/video, moving Prec@1 53.096 Prec@5 80.900
video 5440 done, total 5440/11522, average 0.148 sec/video, moving Prec@1 53.262 Prec@5 81.067
video 5760 done, total 5760/11522, average 0.148 sec/video, moving Prec@1 53.220 Prec@5 80.990
video 6080 done, total 6080/11522, average 0.148 sec/video, moving Prec@1 53.199 Prec@5 81.119
video 6400 done, total 6400/11522, average 0.148 sec/video, moving Prec@1 53.226 Prec@5 81.219
video 6720 done, total 6720/11522, average 0.148 sec/video, moving Prec@1 53.162 Prec@5 81.191
video 7040 done, total 7040/11522, average 0.148 sec/video, moving Prec@1 53.090 Prec@5 81.179
video 7360 done, total 7360/11522, average 0.148 sec/video, moving Prec@1 53.050 Prec@5 81.047
video 7680 done, total 7680/11522, average 0.148 sec/video, moving Prec@1 53.106 Prec@5 81.120
video 8000 done, total 8000/11522, average 0.148 sec/video, moving Prec@1 53.094 Prec@5 81.025
video 8320 done, total 8320/11522, average 0.148 sec/video, moving Prec@1 53.095 Prec@5 81.130
video 8640 done, total 8640/11522, average 0.148 sec/video, moving Prec@1 52.934 Prec@5 81.077
video 8960 done, total 8960/11522, average 0.148 sec/video, moving Prec@1 53.030 Prec@5 81.061
video 9280 done, total 9280/11522, average 0.148 sec/video, moving Prec@1 52.937 Prec@5 81.003
video 9600 done, total 9600/11522, average 0.148 sec/video, moving Prec@1 53.068 Prec@5 81.011
video 9920 done, total 9920/11522, average 0.148 sec/video, moving Prec@1 53.110 Prec@5 80.988
video 10240 done, total 10240/11522, average 0.148 sec/video, moving Prec@1 53.218 Prec@5 81.094
video 10560 done, total 10560/11522, average 0.148 sec/video, moving Prec@1 53.262 Prec@5 81.014
video 10880 done, total 10880/11522, average 0.147 sec/video, moving Prec@1 53.387 Prec@5 81.057
video 11200 done, total 11200/11522, average 0.147 sec/video, moving Prec@1 53.424 Prec@5 81.054
video 11520 done, total 11520/11522, average 0.147 sec/video, moving Prec@1 53.428 Prec@5 81.062
[0.92537313 0.4125     0.35238095 0.68656716 0.36842105 0.52586207
 0.70149254 0.52941176 0.73786408 0.73786408 0.57333333 0.54166667
 0.40425532 0.4516129  0.79617834 0.44       0.24698795 0.2875
 0.42276423 0.4        0.456      0.62068966 0.55555556 0.47058824
 0.4516129  0.33333333 0.47826087 0.36       0.66666667 0.68253968
 0.68181818 0.53846154 0.85185185 0.30769231 0.2173913  0.13888889
 0.8045977  0.80392157 0.10526316 0.66666667 0.72727273 0.64
 0.7        0.69387755 0.73076923 0.63934426 0.47794118 0.41052632
 0.28089888 0.7443609  0.67924528 0.         0.44       0.5
 0.14285714 0.20833333 0.36363636 0.53125    0.         0.71428571
 0.81818182 0.25       0.58       0.13636364 0.16666667 0.5
 0.16666667 0.24242424 0.47887324 0.30769231 0.5483871  0.32142857
 0.5        0.56       0.81666667 0.5483871  0.23529412 0.58823529
 0.31818182 0.85365854 0.08571429 0.23809524 0.33333333 0.21052632
 0.62857143 0.43103448 0.67567568 0.79365079 0.05555556 0.14285714
 0.64864865 0.73170732 0.55813953 0.62962963 0.78504673 0.33870968
 0.03703704 0.16666667 0.59649123 0.5        0.39795918 0.5462963
 0.234375   0.42857143 0.70731707 0.66666667 0.52808989 0.69565217
 0.49122807 0.37362637 0.64705882 0.05882353 0.67567568 0.33333333
 0.66981132 0.46666667 0.29090909 0.78723404 0.31707317 0.625
 0.79545455 0.60784314 0.39090909 0.45       0.27777778 0.48979592
 0.52830189 0.57407407 0.15172414 0.68269231 0.26       0.22222222
 0.31578947 0.33333333 0.7218543  0.46551724 0.07142857 0.16666667
 0.11864407 0.60526316 0.69       0.34883721 0.47727273 0.42372881
 0.54666667 0.50704225 0.62264151 0.19402985 0.625      0.784
 0.6969697  0.22580645 0.36666667 0.85185185 0.64285714 0.21568627
 0.27868852 0.37142857 0.52475248 0.3125     0.5        0.4
 0.34       0.4137931  0.67647059 0.8030303  0.93203883 0.9245283
 0.75925926 0.72972973 0.41904762 0.6618705  0.7080292  0.72368421]
upper bound: 0.507994552098881
-----Evaluation is finished------
Class Accuracy 48.71%
Overall Prec@1 53.43% Prec@5 81.06%
/home/hs/anaconda3/envs/dzl/lib/python3.7/site-packages/numpy/core/_asarray.py:136: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray
  return array(a, dtype, copy=False, order=order, subok=True)

#20201012 MyPA&MyRPA&RGB

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
# python test_models.py something --VAP --batch_size=16 -j=4 --test_crops=1 --test_segments=8,8,8 --weights=pretrained/PAN_RGB_something_resnet50_shift8_blockres_avg_segment8_e50.pth.tar,pretrained/PAN_PA_something_resnet50_shift8_blockres_avg_segment8_e80.pth.tar,pretrained/PAN_RPA_something_resnet50_shift8_blockres_avg_segment8_e80.pth.tar --coeff=2,1,1
$ bash scripts/test/sthv1/3.sh 
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     RGB
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256
            
=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
video number:11522
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     PA
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256
            
=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
video number:11522
video 0 done, total 0/11522, average 0.328 sec/video, moving Prec@1 50.000 Prec@5 87.500
video 320 done, total 320/11522, average 0.055 sec/video, moving Prec@1 56.845 Prec@5 83.036
video 640 done, total 640/11522, average 0.051 sec/video, moving Prec@1 56.555 Prec@5 82.317
video 960 done, total 960/11522, average 0.049 sec/video, moving Prec@1 55.225 Prec@5 80.328
video 1280 done, total 1280/11522, average 0.048 sec/video, moving Prec@1 55.941 Prec@5 80.478
video 1600 done, total 1600/11522, average 0.047 sec/video, moving Prec@1 55.755 Prec@5 80.012
video 1920 done, total 1920/11522, average 0.047 sec/video, moving Prec@1 54.545 Prec@5 79.855
video 2240 done, total 2240/11522, average 0.047 sec/video, moving Prec@1 54.388 Prec@5 79.920
video 2560 done, total 2560/11522, average 0.046 sec/video, moving Prec@1 53.998 Prec@5 80.241
video 2880 done, total 2880/11522, average 0.046 sec/video, moving Prec@1 53.522 Prec@5 80.180
video 3200 done, total 3200/11522, average 0.046 sec/video, moving Prec@1 53.016 Prec@5 80.162
video 3520 done, total 3520/11522, average 0.046 sec/video, moving Prec@1 52.771 Prec@5 80.090
video 3840 done, total 3840/11522, average 0.046 sec/video, moving Prec@1 52.567 Prec@5 79.694
video 4160 done, total 4160/11522, average 0.046 sec/video, moving Prec@1 52.275 Prec@5 79.717
video 4480 done, total 4480/11522, average 0.046 sec/video, moving Prec@1 52.069 Prec@5 79.760
video 4800 done, total 4800/11522, average 0.046 sec/video, moving Prec@1 51.973 Prec@5 79.485
video 5120 done, total 5120/11522, average 0.046 sec/video, moving Prec@1 51.616 Prec@5 79.167
video 5440 done, total 5440/11522, average 0.046 sec/video, moving Prec@1 51.778 Prec@5 79.362
video 5760 done, total 5760/11522, average 0.046 sec/video, moving Prec@1 51.697 Prec@5 79.172
video 6080 done, total 6080/11522, average 0.046 sec/video, moving Prec@1 51.657 Prec@5 79.298
video 6400 done, total 6400/11522, average 0.046 sec/video, moving Prec@1 51.683 Prec@5 79.426
video 6720 done, total 6720/11522, average 0.046 sec/video, moving Prec@1 51.618 Prec@5 79.424
video 7040 done, total 7040/11522, average 0.046 sec/video, moving Prec@1 51.644 Prec@5 79.365
video 7360 done, total 7360/11522, average 0.046 sec/video, moving Prec@1 51.518 Prec@5 79.298
video 7680 done, total 7680/11522, average 0.046 sec/video, moving Prec@1 51.624 Prec@5 79.483
video 8000 done, total 8000/11522, average 0.046 sec/video, moving Prec@1 51.622 Prec@5 79.516
video 8320 done, total 8320/11522, average 0.045 sec/video, moving Prec@1 51.667 Prec@5 79.547
video 8640 done, total 8640/11522, average 0.045 sec/video, moving Prec@1 51.502 Prec@5 79.598
video 8960 done, total 8960/11522, average 0.045 sec/video, moving Prec@1 51.515 Prec@5 79.623
video 9280 done, total 9280/11522, average 0.046 sec/video, moving Prec@1 51.549 Prec@5 79.636
video 9600 done, total 9600/11522, average 0.046 sec/video, moving Prec@1 51.674 Prec@5 79.690
video 9920 done, total 9920/11522, average 0.045 sec/video, moving Prec@1 51.671 Prec@5 79.760
video 10240 done, total 10240/11522, average 0.045 sec/video, moving Prec@1 51.765 Prec@5 79.856
video 10560 done, total 10560/11522, average 0.045 sec/video, moving Prec@1 51.834 Prec@5 79.803
video 10880 done, total 10880/11522, average 0.045 sec/video, moving Prec@1 51.927 Prec@5 79.892
video 11200 done, total 11200/11522, average 0.045 sec/video, moving Prec@1 52.015 Prec@5 79.913
video 11520 done, total 11520/11522, average 0.045 sec/video, moving Prec@1 51.883 Prec@5 79.865
[0.89552239 0.35       0.28571429 0.62686567 0.34210526 0.46551724
 0.68656716 0.41176471 0.69902913 0.66990291 0.36       0.54166667
 0.32978723 0.41935484 0.75159236 0.54       0.24096386 0.275
 0.45528455 0.34666667 0.344      0.5862069  0.40740741 0.45098039
 0.4516129  0.41666667 0.43478261 0.52       0.6969697  0.65079365
 0.59090909 0.57692308 0.83333333 0.24175824 0.32608696 0.16666667
 0.8045977  0.80392157 0.05263158 0.59259259 0.67768595 0.7
 0.71538462 0.65306122 0.71153846 0.60655738 0.45588235 0.44210526
 0.20224719 0.69924812 0.66037736 0.         0.36       0.4
 0.14285714 0.29166667 0.42424242 0.5625     0.08333333 0.74489796
 0.75757576 0.27777778 0.64       0.19090909 0.13333333 0.55555556
 0.14814815 0.22727273 0.42253521 0.34615385 0.5483871  0.28571429
 0.47368421 0.52       0.81666667 0.51612903 0.35294118 0.47058824
 0.36363636 0.7804878  0.08571429 0.16666667 0.4        0.21052632
 0.6        0.34482759 0.62162162 0.68253968 0.33333333 0.10714286
 0.59459459 0.80487805 0.51162791 0.62037037 0.71028037 0.30645161
 0.07407407 0.14285714 0.55555556 0.47058824 0.32653061 0.57407407
 0.21875    0.63265306 0.62195122 0.67619048 0.52808989 0.72463768
 0.49122807 0.34065934 0.58823529 0.05882353 0.68918919 0.38095238
 0.66037736 0.4        0.38181818 0.80851064 0.36585366 0.58333333
 0.84090909 0.58823529 0.39090909 0.5        0.33333333 0.46938776
 0.49056604 0.61111111 0.16551724 0.64423077 0.24       0.25925926
 0.31578947 0.23333333 0.69536424 0.48275862 0.07142857 0.4
 0.20338983 0.61842105 0.66       0.39534884 0.38636364 0.47457627
 0.52       0.45070423 0.61320755 0.17910448 0.546875   0.8
 0.64242424 0.30645161 0.38333333 0.81481481 0.54761905 0.21568627
 0.26229508 0.35714286 0.54455446 0.375      0.48148148 0.35
 0.34       0.4137931  0.73529412 0.74242424 0.9223301  0.86792453
 0.72222222 0.72972973 0.42857143 0.62589928 0.66423358 0.71052632]
upper bound: 0.4917864461284819
-----Evaluation is finished------
Class Accuracy 47.79%
Overall Prec@1 51.88% Prec@5 79.86%
/home/hs/anaconda3/envs/dzl/lib/python3.7/site-packages/numpy/core/_asarray.py:136: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray
  return array(a, dtype, copy=False, order=order, subok=True)

#20201014 RGBo&PAn

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
# python test_models.py something --VAP --batch_size=16 -j=4 --test_crops=1 --test_segments=8,8 --weights=
# pretrained/PAN_RGB_something_resnet50_shift8_blockres_avg_segment8_e50.pth.tar,
# pretrained/PAN_PA_something_resnet50_shift8_blockres_avg_segment8_e80.pth.tar 
$ bash scripts/test/sthv1/Full.sh 
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     RGB
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256
            
=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
video number:11522
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     PA
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256
            
=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
video number:11522
video 0 done, total 0/11522, average 0.249 sec/video, moving Prec@1 50.000 Prec@5 81.250
video 320 done, total 320/11522, average 0.034 sec/video, moving Prec@1 55.655 Prec@5 83.333
video 640 done, total 640/11522, average 0.029 sec/video, moving Prec@1 55.793 Prec@5 81.860
video 960 done, total 960/11522, average 0.027 sec/video, moving Prec@1 54.303 Prec@5 80.328
video 1280 done, total 1280/11522, average 0.026 sec/video, moving Prec@1 54.552 Prec@5 80.633
video 1600 done, total 1600/11522, average 0.026 sec/video, moving Prec@1 54.270 Prec@5 80.012
video 1920 done, total 1920/11522, average 0.025 sec/video, moving Prec@1 53.048 Prec@5 79.597
video 2240 done, total 2240/11522, average 0.025 sec/video, moving Prec@1 53.059 Prec@5 79.654
video 2560 done, total 2560/11522, average 0.025 sec/video, moving Prec@1 52.717 Prec@5 79.697
video 2880 done, total 2880/11522, average 0.025 sec/video, moving Prec@1 52.314 Prec@5 79.731
video 3200 done, total 3200/11522, average 0.025 sec/video, moving Prec@1 52.146 Prec@5 79.633
video 3520 done, total 3520/11522, average 0.024 sec/video, moving Prec@1 51.923 Prec@5 79.525
video 3840 done, total 3840/11522, average 0.024 sec/video, moving Prec@1 51.867 Prec@5 79.175
video 4160 done, total 4160/11522, average 0.024 sec/video, moving Prec@1 51.700 Prec@5 79.119
video 4480 done, total 4480/11522, average 0.024 sec/video, moving Prec@1 51.624 Prec@5 79.181
video 4800 done, total 4800/11522, average 0.024 sec/video, moving Prec@1 51.329 Prec@5 78.841
video 5120 done, total 5120/11522, average 0.024 sec/video, moving Prec@1 50.974 Prec@5 78.583
video 5440 done, total 5440/11522, average 0.024 sec/video, moving Prec@1 51.228 Prec@5 78.757
video 5760 done, total 5760/11522, average 0.024 sec/video, moving Prec@1 51.108 Prec@5 78.636
video 6080 done, total 6080/11522, average 0.024 sec/video, moving Prec@1 51.050 Prec@5 78.773
video 6400 done, total 6400/11522, average 0.024 sec/video, moving Prec@1 51.044 Prec@5 78.959
video 6720 done, total 6720/11522, average 0.024 sec/video, moving Prec@1 50.920 Prec@5 78.934
video 7040 done, total 7040/11522, average 0.024 sec/video, moving Prec@1 50.935 Prec@5 78.784
video 7360 done, total 7360/11522, average 0.024 sec/video, moving Prec@1 50.908 Prec@5 78.783
video 7680 done, total 7680/11522, average 0.024 sec/video, moving Prec@1 51.014 Prec@5 78.950
video 8000 done, total 8000/11522, average 0.024 sec/video, moving Prec@1 50.998 Prec@5 79.004
video 8320 done, total 8320/11522, average 0.024 sec/video, moving Prec@1 50.960 Prec@5 79.007
video 8640 done, total 8640/11522, average 0.024 sec/video, moving Prec@1 50.843 Prec@5 78.963
video 8960 done, total 8960/11522, average 0.024 sec/video, moving Prec@1 50.869 Prec@5 78.966
video 9280 done, total 9280/11522, average 0.024 sec/video, moving Prec@1 50.925 Prec@5 78.959
video 9600 done, total 9600/11522, average 0.024 sec/video, moving Prec@1 51.050 Prec@5 78.973
video 9920 done, total 9920/11522, average 0.024 sec/video, moving Prec@1 51.027 Prec@5 79.006
video 10240 done, total 10240/11522, average 0.024 sec/video, moving Prec@1 51.160 Prec@5 79.105
video 10560 done, total 10560/11522, average 0.024 sec/video, moving Prec@1 51.248 Prec@5 79.085
video 10880 done, total 10880/11522, average 0.024 sec/video, moving Prec@1 51.276 Prec@5 79.157
video 11200 done, total 11200/11522, average 0.024 sec/video, moving Prec@1 51.302 Prec@5 79.173
video 11520 done, total 11520/11522, average 0.024 sec/video, moving Prec@1 51.206 Prec@5 79.170
[0.88059701 0.3375     0.26666667 0.62686567 0.31578947 0.49137931
 0.6641791  0.5        0.69902913 0.68932039 0.38666667 0.55555556
 0.31914894 0.38709677 0.75159236 0.54       0.24096386 0.2625
 0.44715447 0.28       0.336      0.5862069  0.40740741 0.47058824
 0.4516129  0.41666667 0.47826087 0.52       0.75757576 0.65079365
 0.63636364 0.57692308 0.85185185 0.23076923 0.34782609 0.08333333
 0.7816092  0.79411765 0.07894737 0.59259259 0.66942149 0.66
 0.73076923 0.65306122 0.75       0.62295082 0.43382353 0.43157895
 0.17977528 0.69924812 0.58490566 0.         0.32       0.45
 0.14285714 0.20833333 0.45454545 0.5        0.08333333 0.74489796
 0.72727273 0.22222222 0.64       0.17272727 0.13333333 0.50925926
 0.12962963 0.18181818 0.46478873 0.38461538 0.51612903 0.32142857
 0.5        0.56       0.81666667 0.51612903 0.29411765 0.52941176
 0.36363636 0.82926829 0.14285714 0.14285714 0.33333333 0.21052632
 0.65714286 0.36206897 0.62162162 0.71428571 0.33333333 0.10714286
 0.56756757 0.80487805 0.48837209 0.59259259 0.69158879 0.30645161
 0.07407407 0.11904762 0.56140351 0.5        0.32653061 0.57407407
 0.203125   0.65306122 0.69512195 0.65714286 0.50561798 0.72463768
 0.47368421 0.37362637 0.62745098 0.05882353 0.63513514 0.38095238
 0.6509434  0.46666667 0.30909091 0.78723404 0.29268293 0.58333333
 0.80681818 0.58823529 0.4        0.5        0.33333333 0.44897959
 0.50943396 0.57407407 0.16551724 0.60576923 0.24       0.22222222
 0.31578947 0.23333333 0.70198675 0.4137931  0.07142857 0.38333333
 0.18644068 0.57894737 0.68       0.44186047 0.40909091 0.44915254
 0.56       0.42253521 0.62264151 0.1641791  0.4375     0.784
 0.63636364 0.24193548 0.4        0.81481481 0.52380952 0.2745098
 0.2295082  0.42857143 0.51485149 0.34375    0.44444444 0.3
 0.36       0.4137931  0.69607843 0.75757576 0.9223301  0.8490566
 0.69444444 0.75675676 0.41904762 0.63309353 0.64233577 0.68421053]
upper bound: 0.4885991836244713
-----Evaluation is finished------
Class Accuracy 47.34%
Overall Prec@1 51.21% Prec@5 79.17%
/home/hs/anaconda3/envs/dzl/lib/python3.7/site-packages/numpy/core/_asarray.py:136: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray
  return array(a, dtype, copy=False, order=order, subok=True)

#20201014 RGBo&PAo&PAn

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
# python test_models.py something --VAP --batch_size=16 -j=4 --test_crops=1 --test_segments=8,8,8 --weights=
# pretrained/PAN_RGB_something_resnet50_shift8_blockres_avg_segment8_e50.pth.tar,
# pretrained/PAN_PA_something_resnet50_shift8_blockres_avg_segment8_e80.pth.tar,
# o/PAN_PA_something_resnet50_shift8_blockres_avg_segment8_e80.pth.tar --coeff=2,1,1
$ bash scripts/test/sthv1/3-old-new.sh 
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     RGB
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256
            
=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
video number:11522
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     PA
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256
            
=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
video number:11522
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     PA
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256
            
=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
video number:11522
video 0 done, total 0/11522, average 0.323 sec/video, moving Prec@1 50.000 Prec@5 81.250
video 320 done, total 320/11522, average 0.056 sec/video, moving Prec@1 53.869 Prec@5 82.440
video 640 done, total 640/11522, average 0.050 sec/video, moving Prec@1 54.116 Prec@5 82.012
video 960 done, total 960/11522, average 0.048 sec/video, moving Prec@1 53.074 Prec@5 80.738
video 1280 done, total 1280/11522, average 0.048 sec/video, moving Prec@1 53.935 Prec@5 81.019
video 1600 done, total 1600/11522, average 0.047 sec/video, moving Prec@1 53.713 Prec@5 80.507
video 1920 done, total 1920/11522, average 0.047 sec/video, moving Prec@1 52.841 Prec@5 80.269
video 2240 done, total 2240/11522, average 0.046 sec/video, moving Prec@1 52.881 Prec@5 80.408
video 2560 done, total 2560/11522, average 0.046 sec/video, moving Prec@1 52.911 Prec@5 80.551
video 2880 done, total 2880/11522, average 0.046 sec/video, moving Prec@1 52.521 Prec@5 80.456
video 3200 done, total 3200/11522, average 0.046 sec/video, moving Prec@1 52.208 Prec@5 80.442
video 3520 done, total 3520/11522, average 0.045 sec/video, moving Prec@1 51.951 Prec@5 80.373
video 3840 done, total 3840/11522, average 0.045 sec/video, moving Prec@1 51.893 Prec@5 80.031
video 4160 done, total 4160/11522, average 0.045 sec/video, moving Prec@1 51.652 Prec@5 79.909
video 4480 done, total 4480/11522, average 0.045 sec/video, moving Prec@1 51.624 Prec@5 79.982
video 4800 done, total 4800/11522, average 0.045 sec/video, moving Prec@1 51.433 Prec@5 79.713
video 5120 done, total 5120/11522, average 0.045 sec/video, moving Prec@1 51.168 Prec@5 79.459
video 5440 done, total 5440/11522, average 0.045 sec/video, moving Prec@1 51.430 Prec@5 79.619
video 5760 done, total 5760/11522, average 0.045 sec/video, moving Prec@1 51.333 Prec@5 79.519
video 6080 done, total 6080/11522, average 0.045 sec/video, moving Prec@1 51.345 Prec@5 79.544
video 6400 done, total 6400/11522, average 0.045 sec/video, moving Prec@1 51.418 Prec@5 79.723
video 6720 done, total 6720/11522, average 0.045 sec/video, moving Prec@1 51.232 Prec@5 79.706
video 7040 done, total 7040/11522, average 0.045 sec/video, moving Prec@1 51.247 Prec@5 79.592
video 7360 done, total 7360/11522, average 0.045 sec/video, moving Prec@1 51.166 Prec@5 79.515
video 7680 done, total 7680/11522, average 0.045 sec/video, moving Prec@1 51.312 Prec@5 79.652
video 8000 done, total 8000/11522, average 0.045 sec/video, moving Prec@1 51.322 Prec@5 79.703
video 8320 done, total 8320/11522, average 0.045 sec/video, moving Prec@1 51.308 Prec@5 79.690
video 8640 done, total 8640/11522, average 0.045 sec/video, moving Prec@1 51.178 Prec@5 79.644
video 8960 done, total 8960/11522, average 0.045 sec/video, moving Prec@1 51.203 Prec@5 79.635
video 9280 done, total 9280/11522, average 0.045 sec/video, moving Prec@1 51.216 Prec@5 79.615
video 9600 done, total 9600/11522, average 0.045 sec/video, moving Prec@1 51.373 Prec@5 79.638
video 9920 done, total 9920/11522, average 0.045 sec/video, moving Prec@1 51.318 Prec@5 79.670
video 10240 done, total 10240/11522, average 0.045 sec/video, moving Prec@1 51.424 Prec@5 79.797
video 10560 done, total 10560/11522, average 0.045 sec/video, moving Prec@1 51.522 Prec@5 79.784
video 10880 done, total 10880/11522, average 0.045 sec/video, moving Prec@1 51.597 Prec@5 79.837
video 11200 done, total 11200/11522, average 0.045 sec/video, moving Prec@1 51.658 Prec@5 79.850
video 11520 done, total 11520/11522, average 0.044 sec/video, moving Prec@1 51.554 Prec@5 79.847
[0.89552239 0.3375     0.24761905 0.67164179 0.34210526 0.50862069
 0.70895522 0.44117647 0.66019417 0.7184466  0.4        0.48611111
 0.35106383 0.35483871 0.76433121 0.52       0.23493976 0.2875
 0.45528455 0.36       0.336      0.62068966 0.48148148 0.49019608
 0.51612903 0.41666667 0.47826087 0.4        0.72727273 0.65079365
 0.59090909 0.57692308 0.87037037 0.25274725 0.2826087  0.16666667
 0.82758621 0.80392157 0.05263158 0.55555556 0.66942149 0.66
 0.73076923 0.67346939 0.76923077 0.63934426 0.45588235 0.42105263
 0.16853933 0.69172932 0.63207547 0.         0.36       0.45
 0.14285714 0.29166667 0.42424242 0.5        0.         0.75510204
 0.75757576 0.13888889 0.64       0.16363636 0.1        0.49074074
 0.18518519 0.21212121 0.46478873 0.30769231 0.5483871  0.28571429
 0.5        0.56       0.83333333 0.4516129  0.35294118 0.58823529
 0.27272727 0.80487805 0.08571429 0.14285714 0.36666667 0.15789474
 0.65714286 0.37931034 0.62162162 0.68253968 0.33333333 0.07142857
 0.54054054 0.75609756 0.53488372 0.57407407 0.6728972  0.29032258
 0.07407407 0.11904762 0.55555556 0.48529412 0.34693878 0.53703704
 0.203125   0.55102041 0.7195122  0.64761905 0.50561798 0.73913043
 0.47368421 0.37362637 0.60784314 0.05882353 0.64864865 0.33333333
 0.62264151 0.46666667 0.34545455 0.77659574 0.34146341 0.60416667
 0.80681818 0.58823529 0.38181818 0.5        0.33333333 0.46938776
 0.47169811 0.61111111 0.15862069 0.63461538 0.24       0.22222222
 0.23684211 0.33333333 0.70198675 0.4137931  0.         0.4
 0.16949153 0.55263158 0.66       0.44186047 0.38636364 0.45762712
 0.56       0.45774648 0.63207547 0.17910448 0.484375   0.808
 0.65454545 0.30645161 0.41666667 0.7962963  0.57142857 0.2745098
 0.26229508 0.34285714 0.5049505  0.34375    0.42592593 0.325
 0.32       0.4137931  0.69607843 0.77272727 0.9223301  0.86792453
 0.72222222 0.75675676 0.42857143 0.64028777 0.66423358 0.73684211]
upper bound: 0.49272722679828856
-----Evaluation is finished------
Class Accuracy 47.36%
Overall Prec@1 51.55% Prec@5 79.85%
/home/hs/anaconda3/envs/dzl/lib/python3.7/site-packages/numpy/core/_asarray.py:136: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray
  return array(a, dtype, copy=False, order=order, subok=True)

#20201014 RGBo&PAn&LPAn

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
# python test_models.py something --VAP --batch_size=16 -j=4 --test_crops=1 --test_segments=8,8,8 --weights=
# pretrained/PAN_RGB_something_resnet50_shift8_blockres_avg_segment8_e50.pth.tar,
# pretrained/PAN_PA_something_resnet50_shift8_blockres_avg_segment8_e80.pth.tar,
# pretrained/PAN_LPA_something_resnet50_shift8_blockres_avg_segment8_e80.pth.tar --coeff=2,1,1
$ bash scripts/test/sthv1/3-lpa-pa.sh 
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     RGB
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256
            
=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
video number:11522
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     PA
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256
            
=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
video number:11522
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     PA
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256
            
=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
video number:11522
video 0 done, total 0/11522, average 0.302 sec/video, moving Prec@1 43.750 Prec@5 81.250
video 320 done, total 320/11522, average 0.055 sec/video, moving Prec@1 54.464 Prec@5 81.845
video 640 done, total 640/11522, average 0.050 sec/video, moving Prec@1 54.878 Prec@5 82.317
video 960 done, total 960/11522, average 0.049 sec/video, moving Prec@1 53.074 Prec@5 80.533
video 1280 done, total 1280/11522, average 0.048 sec/video, moving Prec@1 53.704 Prec@5 81.250
video 1600 done, total 1600/11522, average 0.048 sec/video, moving Prec@1 53.837 Prec@5 80.817
video 1920 done, total 1920/11522, average 0.048 sec/video, moving Prec@1 52.738 Prec@5 80.475
video 2240 done, total 2240/11522, average 0.047 sec/video, moving Prec@1 52.793 Prec@5 80.496
video 2560 done, total 2560/11522, average 0.047 sec/video, moving Prec@1 52.484 Prec@5 80.668
video 2880 done, total 2880/11522, average 0.047 sec/video, moving Prec@1 52.141 Prec@5 80.594
video 3200 done, total 3200/11522, average 0.047 sec/video, moving Prec@1 51.835 Prec@5 80.597
video 3520 done, total 3520/11522, average 0.047 sec/video, moving Prec@1 51.753 Prec@5 80.571
video 3840 done, total 3840/11522, average 0.047 sec/video, moving Prec@1 51.738 Prec@5 80.083
video 4160 done, total 4160/11522, average 0.047 sec/video, moving Prec@1 51.269 Prec@5 79.981
video 4480 done, total 4480/11522, average 0.047 sec/video, moving Prec@1 51.179 Prec@5 80.027
video 4800 done, total 4800/11522, average 0.047 sec/video, moving Prec@1 50.976 Prec@5 79.610
video 5120 done, total 5120/11522, average 0.047 sec/video, moving Prec@1 50.662 Prec@5 79.400
video 5440 done, total 5440/11522, average 0.047 sec/video, moving Prec@1 50.990 Prec@5 79.619
video 5760 done, total 5760/11522, average 0.047 sec/video, moving Prec@1 50.952 Prec@5 79.467
video 6080 done, total 6080/11522, average 0.047 sec/video, moving Prec@1 50.984 Prec@5 79.593
video 6400 done, total 6400/11522, average 0.047 sec/video, moving Prec@1 51.044 Prec@5 79.660
video 6720 done, total 6720/11522, average 0.047 sec/video, moving Prec@1 50.965 Prec@5 79.572
video 7040 done, total 7040/11522, average 0.047 sec/video, moving Prec@1 50.950 Prec@5 79.521
video 7360 done, total 7360/11522, average 0.047 sec/video, moving Prec@1 50.935 Prec@5 79.420
video 7680 done, total 7680/11522, average 0.047 sec/video, moving Prec@1 51.117 Prec@5 79.561
video 8000 done, total 8000/11522, average 0.046 sec/video, moving Prec@1 51.160 Prec@5 79.591
video 8320 done, total 8320/11522, average 0.046 sec/video, moving Prec@1 51.188 Prec@5 79.607
video 8640 done, total 8640/11522, average 0.046 sec/video, moving Prec@1 51.051 Prec@5 79.552
video 8960 done, total 8960/11522, average 0.046 sec/video, moving Prec@1 51.081 Prec@5 79.557
video 9280 done, total 9280/11522, average 0.046 sec/video, moving Prec@1 51.108 Prec@5 79.518
video 9600 done, total 9600/11522, average 0.046 sec/video, moving Prec@1 51.258 Prec@5 79.545
video 9920 done, total 9920/11522, average 0.046 sec/video, moving Prec@1 51.178 Prec@5 79.539
video 10240 done, total 10240/11522, average 0.046 sec/video, moving Prec@1 51.297 Prec@5 79.631
video 10560 done, total 10560/11522, average 0.046 sec/video, moving Prec@1 51.466 Prec@5 79.605
video 10880 done, total 10880/11522, average 0.046 sec/video, moving Prec@1 51.523 Prec@5 79.653
video 11200 done, total 11200/11522, average 0.046 sec/video, moving Prec@1 51.632 Prec@5 79.690
video 11520 done, total 11520/11522, average 0.046 sec/video, moving Prec@1 51.562 Prec@5 79.665
[0.88059701 0.3375     0.24761905 0.65671642 0.28947368 0.46551724
 0.68656716 0.5        0.66990291 0.70873786 0.38666667 0.52777778
 0.29787234 0.35483871 0.76433121 0.52       0.21084337 0.2875
 0.44715447 0.32       0.352      0.62068966 0.38888889 0.49019608
 0.51612903 0.41666667 0.47826087 0.56       0.78787879 0.68253968
 0.63636364 0.57692308 0.87037037 0.21978022 0.2826087  0.19444444
 0.81609195 0.80392157 0.07894737 0.66666667 0.67768595 0.66
 0.71538462 0.69387755 0.75       0.60655738 0.41911765 0.43157895
 0.19101124 0.71428571 0.63207547 0.         0.4        0.45
 0.14285714 0.25       0.48484848 0.5        0.08333333 0.7755102
 0.72727273 0.16666667 0.62       0.2        0.13333333 0.51851852
 0.16666667 0.22727273 0.47887324 0.34615385 0.5483871  0.32142857
 0.5        0.52       0.81666667 0.5483871  0.35294118 0.52941176
 0.40909091 0.80487805 0.11428571 0.14285714 0.36666667 0.21052632
 0.6        0.37931034 0.62162162 0.68253968 0.16666667 0.08928571
 0.59459459 0.80487805 0.53488372 0.57407407 0.71028037 0.29032258
 0.07407407 0.16666667 0.5497076  0.47058824 0.32653061 0.55555556
 0.1875     0.6122449  0.68292683 0.64761905 0.53932584 0.72463768
 0.47368421 0.35164835 0.58823529 0.05882353 0.62162162 0.38095238
 0.66981132 0.5        0.32727273 0.77659574 0.34146341 0.58333333
 0.82954545 0.56862745 0.39090909 0.5        0.33333333 0.46938776
 0.47169811 0.57407407 0.15172414 0.63461538 0.26       0.25925926
 0.26315789 0.33333333 0.68874172 0.39655172 0.14285714 0.31666667
 0.18644068 0.56578947 0.66       0.44186047 0.36363636 0.47457627
 0.58666667 0.45070423 0.61320755 0.14925373 0.46875    0.8
 0.66666667 0.24193548 0.38333333 0.7962963  0.5        0.29411765
 0.27868852 0.42857143 0.48514851 0.34375    0.44444444 0.35
 0.34       0.44827586 0.70588235 0.74242424 0.9223301  0.85849057
 0.71296296 0.75675676 0.44761905 0.65467626 0.65693431 0.71052632]
upper bound: 0.49360881004950025
-----Evaluation is finished------
Class Accuracy 47.77%
Overall Prec@1 51.56% Prec@5 79.66%
/home/hs/anaconda3/envs/dzl/lib/python3.7/site-packages/numpy/core/_asarray.py:136: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray
  return array(a, dtype, copy=False, order=order, subok=True)

#20201014 RGBo&RPAn&LPAn

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
# python test_models.py something --VAP --batch_size=16 -j=4 --test_crops=1 --test_segments=8,8,8 --weights=
# pretrained/PAN_RGB_something_resnet50_shift8_blockres_avg_segment8_e50.pth.tar,
# pretrained/PAN_RPA_something_resnet50_shift8_blockres_avg_segment8_e80.pth.tar,
# pretrained/PAN_LPA_something_resnet50_shift8_blockres_avg_segment8_e80.pth.tar --coeff=2,1,1
$ bash scripts/test/sthv1/3-lpa-rpa.sh 
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     RGB
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256
            
=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
video number:11522
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     PA
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256
            
=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
video number:11522
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     PA
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256
            
=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
video number:11522
video 0 done, total 0/11522, average 0.289 sec/video, moving Prec@1 43.750 Prec@5 87.500
video 320 done, total 320/11522, average 0.054 sec/video, moving Prec@1 55.357 Prec@5 83.333
video 640 done, total 640/11522, average 0.050 sec/video, moving Prec@1 56.250 Prec@5 82.774
video 960 done, total 960/11522, average 0.048 sec/video, moving Prec@1 54.098 Prec@5 80.840
video 1280 done, total 1280/11522, average 0.047 sec/video, moving Prec@1 54.707 Prec@5 81.096
video 1600 done, total 1600/11522, average 0.047 sec/video, moving Prec@1 54.517 Prec@5 80.693
video 1920 done, total 1920/11522, average 0.047 sec/video, moving Prec@1 53.409 Prec@5 80.269
video 2240 done, total 2240/11522, average 0.047 sec/video, moving Prec@1 53.369 Prec@5 80.363
video 2560 done, total 2560/11522, average 0.047 sec/video, moving Prec@1 53.028 Prec@5 80.474
video 2880 done, total 2880/11522, average 0.047 sec/video, moving Prec@1 52.624 Prec@5 80.421
video 3200 done, total 3200/11522, average 0.047 sec/video, moving Prec@1 52.208 Prec@5 80.410
video 3520 done, total 3520/11522, average 0.047 sec/video, moving Prec@1 52.291 Prec@5 80.373
video 3840 done, total 3840/11522, average 0.047 sec/video, moving Prec@1 52.127 Prec@5 79.850
video 4160 done, total 4160/11522, average 0.047 sec/video, moving Prec@1 51.580 Prec@5 79.765
video 4480 done, total 4480/11522, average 0.047 sec/video, moving Prec@1 51.379 Prec@5 79.782
video 4800 done, total 4800/11522, average 0.047 sec/video, moving Prec@1 51.204 Prec@5 79.402
video 5120 done, total 5120/11522, average 0.047 sec/video, moving Prec@1 50.896 Prec@5 79.167
video 5440 done, total 5440/11522, average 0.047 sec/video, moving Prec@1 51.045 Prec@5 79.417
video 5760 done, total 5760/11522, average 0.047 sec/video, moving Prec@1 50.952 Prec@5 79.259
video 6080 done, total 6080/11522, average 0.047 sec/video, moving Prec@1 51.050 Prec@5 79.446
video 6400 done, total 6400/11522, average 0.047 sec/video, moving Prec@1 51.122 Prec@5 79.551
video 6720 done, total 6720/11522, average 0.047 sec/video, moving Prec@1 51.024 Prec@5 79.543
video 7040 done, total 7040/11522, average 0.047 sec/video, moving Prec@1 51.006 Prec@5 79.493
video 7360 done, total 7360/11522, average 0.047 sec/video, moving Prec@1 51.003 Prec@5 79.501
video 7680 done, total 7680/11522, average 0.047 sec/video, moving Prec@1 51.156 Prec@5 79.665
video 8000 done, total 8000/11522, average 0.047 sec/video, moving Prec@1 51.135 Prec@5 79.678
video 8320 done, total 8320/11522, average 0.047 sec/video, moving Prec@1 51.164 Prec@5 79.750
video 8640 done, total 8640/11522, average 0.047 sec/video, moving Prec@1 50.970 Prec@5 79.748
video 8960 done, total 8960/11522, average 0.046 sec/video, moving Prec@1 50.925 Prec@5 79.779
video 9280 done, total 9280/11522, average 0.046 sec/video, moving Prec@1 50.957 Prec@5 79.755
video 9600 done, total 9600/11522, average 0.047 sec/video, moving Prec@1 51.102 Prec@5 79.815
video 9920 done, total 9920/11522, average 0.047 sec/video, moving Prec@1 51.057 Prec@5 79.821
video 10240 done, total 10240/11522, average 0.046 sec/video, moving Prec@1 51.180 Prec@5 79.914
video 10560 done, total 10560/11522, average 0.046 sec/video, moving Prec@1 51.276 Prec@5 79.898
video 10880 done, total 10880/11522, average 0.046 sec/video, moving Prec@1 51.358 Prec@5 79.965
video 11200 done, total 11200/11522, average 0.046 sec/video, moving Prec@1 51.462 Prec@5 79.993
video 11520 done, total 11520/11522, average 0.046 sec/video, moving Prec@1 51.380 Prec@5 79.934
[0.89552239 0.3375     0.25714286 0.62686567 0.31578947 0.44827586
 0.70149254 0.44117647 0.66019417 0.69902913 0.38666667 0.51388889
 0.30851064 0.41935484 0.74522293 0.5        0.21084337 0.2625
 0.44715447 0.4        0.328      0.60344828 0.35185185 0.43137255
 0.4516129  0.41666667 0.47826087 0.56       0.75757576 0.66666667
 0.72727273 0.57692308 0.83333333 0.21978022 0.30434783 0.19444444
 0.8045977  0.80392157 0.07894737 0.66666667 0.66942149 0.62
 0.7        0.67346939 0.75       0.60655738 0.44852941 0.38947368
 0.15730337 0.70676692 0.6509434  0.         0.36       0.45
 0.14285714 0.25       0.42424242 0.53125    0.         0.78571429
 0.75757576 0.25       0.64       0.22727273 0.13333333 0.50925926
 0.16666667 0.25757576 0.45070423 0.42307692 0.5483871  0.28571429
 0.47368421 0.52       0.76666667 0.5483871  0.35294118 0.47058824
 0.36363636 0.80487805 0.08571429 0.14285714 0.4        0.15789474
 0.54285714 0.34482759 0.60810811 0.66666667 0.22222222 0.08928571
 0.56756757 0.80487805 0.58139535 0.56481481 0.68224299 0.29032258
 0.07407407 0.11904762 0.56140351 0.44117647 0.32653061 0.55555556
 0.21875    0.63265306 0.64634146 0.62857143 0.5505618  0.72463768
 0.47368421 0.32967033 0.58823529 0.05882353 0.64864865 0.38095238
 0.66037736 0.4        0.38181818 0.78723404 0.36585366 0.5625
 0.80681818 0.60784314 0.39090909 0.5        0.30555556 0.46938776
 0.47169811 0.62962963 0.14482759 0.61538462 0.3        0.25925926
 0.26315789 0.26666667 0.68211921 0.44827586 0.07142857 0.31666667
 0.16949153 0.59210526 0.65       0.41860465 0.40909091 0.46610169
 0.6        0.43661972 0.61320755 0.11940299 0.5625     0.792
 0.67272727 0.25806452 0.36666667 0.81481481 0.57142857 0.19607843
 0.27868852 0.38571429 0.52475248 0.375      0.5        0.4
 0.32       0.44827586 0.71568627 0.75757576 0.9223301  0.87735849
 0.73148148 0.75675676 0.46666667 0.64028777 0.64233577 0.68421053]
upper bound: 0.49090826225102707
-----Evaluation is finished------
Class Accuracy 47.50%
Overall Prec@1 51.38% Prec@5 79.93%
/home/hs/anaconda3/envs/dzl/lib/python3.7/site-packages/numpy/core/_asarray.py:136: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray
  return array(a, dtype, copy=False, order=order, subok=True)

#20201014 RGBo&LPAn

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
# python test_models.py something --VAP --batch_size=16 -j=4 --test_crops=1 --test_segments=8,8 --weights=
# pretrained/PAN_RGB_something_resnet50_shift8_blockres_avg_segment8_e50.pth.tar,
# pretrained/PAN_LPA_something_resnet50_shift8_blockres_avg_segment8_e80.pth.tar 
$ bash scripts/test/sthv1/2-lpa.sh 
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     RGB
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256
            
=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
video number:11522
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     PA
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256
            
=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
video number:11522
video 0 done, total 0/11522, average 0.237 sec/video, moving Prec@1 43.750 Prec@5 81.250
video 320 done, total 320/11522, average 0.034 sec/video, moving Prec@1 51.786 Prec@5 83.036
video 640 done, total 640/11522, average 0.029 sec/video, moving Prec@1 53.354 Prec@5 82.470
video 960 done, total 960/11522, average 0.027 sec/video, moving Prec@1 51.742 Prec@5 79.918
video 1280 done, total 1280/11522, average 0.026 sec/video, moving Prec@1 52.469 Prec@5 80.247
video 1600 done, total 1600/11522, average 0.026 sec/video, moving Prec@1 52.413 Prec@5 79.950
video 1920 done, total 1920/11522, average 0.025 sec/video, moving Prec@1 51.601 Prec@5 79.804
video 2240 done, total 2240/11522, average 0.025 sec/video, moving Prec@1 51.817 Prec@5 79.876
video 2560 done, total 2560/11522, average 0.025 sec/video, moving Prec@1 51.708 Prec@5 79.852
video 2880 done, total 2880/11522, average 0.025 sec/video, moving Prec@1 51.450 Prec@5 79.869
video 3200 done, total 3200/11522, average 0.025 sec/video, moving Prec@1 50.964 Prec@5 79.851
video 3520 done, total 3520/11522, average 0.025 sec/video, moving Prec@1 50.962 Prec@5 79.808
video 3840 done, total 3840/11522, average 0.025 sec/video, moving Prec@1 51.011 Prec@5 79.357
video 4160 done, total 4160/11522, average 0.024 sec/video, moving Prec@1 50.694 Prec@5 79.119
video 4480 done, total 4480/11522, average 0.024 sec/video, moving Prec@1 50.556 Prec@5 79.070
video 4800 done, total 4800/11522, average 0.024 sec/video, moving Prec@1 50.374 Prec@5 78.675
video 5120 done, total 5120/11522, average 0.024 sec/video, moving Prec@1 50.117 Prec@5 78.427
video 5440 done, total 5440/11522, average 0.024 sec/video, moving Prec@1 50.385 Prec@5 78.666
video 5760 done, total 5760/11522, average 0.024 sec/video, moving Prec@1 50.346 Prec@5 78.532
video 6080 done, total 6080/11522, average 0.024 sec/video, moving Prec@1 50.476 Prec@5 78.724
video 6400 done, total 6400/11522, average 0.024 sec/video, moving Prec@1 50.530 Prec@5 78.803
video 6720 done, total 6720/11522, average 0.024 sec/video, moving Prec@1 50.386 Prec@5 78.711
video 7040 done, total 7040/11522, average 0.024 sec/video, moving Prec@1 50.368 Prec@5 78.713
video 7360 done, total 7360/11522, average 0.024 sec/video, moving Prec@1 50.325 Prec@5 78.633
video 7680 done, total 7680/11522, average 0.024 sec/video, moving Prec@1 50.533 Prec@5 78.833
video 8000 done, total 8000/11522, average 0.024 sec/video, moving Prec@1 50.487 Prec@5 78.855
video 8320 done, total 8320/11522, average 0.024 sec/video, moving Prec@1 50.504 Prec@5 78.911
video 8640 done, total 8640/11522, average 0.024 sec/video, moving Prec@1 50.335 Prec@5 78.847
video 8960 done, total 8960/11522, average 0.024 sec/video, moving Prec@1 50.267 Prec@5 78.855
video 9280 done, total 9280/11522, average 0.024 sec/video, moving Prec@1 50.258 Prec@5 78.840
video 9600 done, total 9600/11522, average 0.024 sec/video, moving Prec@1 50.374 Prec@5 78.869
video 9920 done, total 9920/11522, average 0.024 sec/video, moving Prec@1 50.312 Prec@5 78.845
video 10240 done, total 10240/11522, average 0.024 sec/video, moving Prec@1 50.468 Prec@5 78.939
video 10560 done, total 10560/11522, average 0.024 sec/video, moving Prec@1 50.605 Prec@5 78.896
video 10880 done, total 10880/11522, average 0.024 sec/video, moving Prec@1 50.698 Prec@5 78.956
video 11200 done, total 11200/11522, average 0.024 sec/video, moving Prec@1 50.811 Prec@5 79.003
video 11520 done, total 11520/11522, average 0.024 sec/video, moving Prec@1 50.790 Prec@5 78.979
[0.91044776 0.3875     0.25714286 0.62686567 0.31578947 0.4137931
 0.69402985 0.52941176 0.65048544 0.72815534 0.4        0.5
 0.28723404 0.41935484 0.75796178 0.46       0.19277108 0.275
 0.45528455 0.36       0.344      0.5862069  0.31481481 0.49019608
 0.48387097 0.41666667 0.47826087 0.52       0.75757576 0.68253968
 0.72727273 0.61538462 0.83333333 0.20879121 0.26086957 0.16666667
 0.81609195 0.80392157 0.05263158 0.66666667 0.66942149 0.64
 0.71538462 0.67346939 0.73076923 0.59016393 0.41176471 0.42105263
 0.15730337 0.69172932 0.63207547 0.         0.4        0.45
 0.14285714 0.20833333 0.36363636 0.46875    0.         0.81632653
 0.66666667 0.16666667 0.62       0.22727273 0.16666667 0.50925926
 0.16666667 0.21212121 0.45070423 0.46153846 0.5483871  0.35714286
 0.5        0.52       0.78333333 0.5483871  0.41176471 0.52941176
 0.36363636 0.82926829 0.11428571 0.11904762 0.36666667 0.15789474
 0.54285714 0.32758621 0.60810811 0.6984127  0.22222222 0.07142857
 0.54054054 0.70731707 0.55813953 0.5462963  0.68224299 0.24193548
 0.07407407 0.19047619 0.54385965 0.39705882 0.33673469 0.53703704
 0.203125   0.59183673 0.62195122 0.63809524 0.56179775 0.73913043
 0.43859649 0.32967033 0.56862745 0.05882353 0.58108108 0.42857143
 0.64150943 0.43333333 0.34545455 0.78723404 0.3902439  0.58333333
 0.81818182 0.58823529 0.39090909 0.5        0.33333333 0.44897959
 0.43396226 0.55555556 0.14482759 0.63461538 0.26       0.18518519
 0.28947368 0.3        0.66887417 0.39655172 0.07142857 0.21666667
 0.13559322 0.59210526 0.62       0.41860465 0.43181818 0.46610169
 0.6        0.42957746 0.61320755 0.11940299 0.484375   0.776
 0.67878788 0.22580645 0.36666667 0.83333333 0.57142857 0.2745098
 0.26229508 0.42857143 0.4950495  0.34375    0.44444444 0.325
 0.26       0.4137931  0.7254902  0.75757576 0.9223301  0.8490566
 0.72222222 0.72972973 0.48571429 0.65467626 0.65693431 0.69736842]
upper bound: 0.4877900527936639
-----Evaluation is finished------
Class Accuracy 46.92%
Overall Prec@1 50.79% Prec@5 78.98%
/home/hs/anaconda3/envs/dzl/lib/python3.7/site-packages/numpy/core/_asarray.py:136: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray
  return array(a, dtype, copy=False, order=order, subok=True)

#20201019 PA with new conv & RGB

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
$ bash scripts/test/sthv1/Full.sh 
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     RGB
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256

=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
video number:11522
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.    
    PAN Configurations:
        input_modality:     PA
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256

=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
video number:11522
video 0 done, total 0/11522, average 0.246 sec/video, moving Prec@1 43.750 Prec@5 87.500
video 320 done, total 320/11522, average 0.034 sec/video, moving Prec@1 54.167 Prec@5 83.036
video 640 done, total 640/11522, average 0.029 sec/video, moving Prec@1 54.421 Prec@5 82.317
video 960 done, total 960/11522, average 0.028 sec/video, moving Prec@1 52.459 Prec@5 80.328
video 1280 done, total 1280/11522, average 0.027 sec/video, moving Prec@1 53.472 Prec@5 80.633
video 1600 done, total 1600/11522, average 0.026 sec/video, moving Prec@1 53.218 Prec@5 80.074
video 1920 done, total 1920/11522, average 0.026 sec/video, moving Prec@1 52.531 Prec@5 79.907
video 2240 done, total 2240/11522, average 0.026 sec/video, moving Prec@1 52.482 Prec@5 79.699
video 2560 done, total 2560/11522, average 0.025 sec/video, moving Prec@1 52.019 Prec@5 79.658
video 2880 done, total 2880/11522, average 0.025 sec/video, moving Prec@1 51.623 Prec@5 79.489
video 3200 done, total 3200/11522, average 0.025 sec/video, moving Prec@1 51.057 Prec@5 79.509
video 3520 done, total 3520/11522, average 0.025 sec/video, moving Prec@1 50.990 Prec@5 79.525
video 3840 done, total 3840/11522, average 0.025 sec/video, moving Prec@1 51.011 Prec@5 79.149
video 4160 done, total 4160/11522, average 0.025 sec/video, moving Prec@1 50.766 Prec@5 79.143
video 4480 done, total 4480/11522, average 0.025 sec/video, moving Prec@1 50.712 Prec@5 79.248
video 4800 done, total 4800/11522, average 0.025 sec/video, moving Prec@1 50.478 Prec@5 78.987
video 5120 done, total 5120/11522, average 0.025 sec/video, moving Prec@1 50.253 Prec@5 78.738
video 5440 done, total 5440/11522, average 0.025 sec/video, moving Prec@1 50.293 Prec@5 78.867
video 5760 done, total 5760/11522, average 0.025 sec/video, moving Prec@1 50.381 Prec@5 78.722
video 6080 done, total 6080/11522, average 0.025 sec/video, moving Prec@1 50.525 Prec@5 78.822
video 6400 done, total 6400/11522, average 0.025 sec/video, moving Prec@1 50.592 Prec@5 79.037
video 6720 done, total 6720/11522, average 0.025 sec/video, moving Prec@1 50.460 Prec@5 78.934
video 7040 done, total 7040/11522, average 0.025 sec/video, moving Prec@1 50.468 Prec@5 78.883
video 7360 done, total 7360/11522, average 0.025 sec/video, moving Prec@1 50.461 Prec@5 78.823
video 7680 done, total 7680/11522, average 0.024 sec/video, moving Prec@1 50.572 Prec@5 79.002
video 8000 done, total 8000/11522, average 0.024 sec/video, moving Prec@1 50.474 Prec@5 79.029
video 8320 done, total 8320/11522, average 0.024 sec/video, moving Prec@1 50.528 Prec@5 79.031
video 8640 done, total 8640/11522, average 0.024 sec/video, moving Prec@1 50.393 Prec@5 78.986
video 8960 done, total 8960/11522, average 0.024 sec/video, moving Prec@1 50.390 Prec@5 79.022
video 9280 done, total 9280/11522, average 0.024 sec/video, moving Prec@1 50.333 Prec@5 78.926
video 9600 done, total 9600/11522, average 0.024 sec/video, moving Prec@1 50.499 Prec@5 78.952
video 9920 done, total 9920/11522, average 0.024 sec/video, moving Prec@1 50.543 Prec@5 78.975
video 10240 done, total 10240/11522, average 0.024 sec/video, moving Prec@1 50.644 Prec@5 79.134
video 10560 done, total 10560/11522, average 0.024 sec/video, moving Prec@1 50.766 Prec@5 79.113
video 10880 done, total 10880/11522, average 0.024 sec/video, moving Prec@1 50.835 Prec@5 79.176
video 11200 done, total 11200/11522, average 0.024 sec/video, moving Prec@1 50.945 Prec@5 79.182
video 11520 done, total 11520/11522, average 0.024 sec/video, moving Prec@1 50.920 Prec@5 79.101
[0.89552239 0.3375     0.26666667 0.62686567 0.34210526 0.45689655
 0.70895522 0.52941176 0.61165049 0.69902913 0.44       0.51388889
 0.29787234 0.38709677 0.74522293 0.46       0.18674699 0.275
 0.46341463 0.38666667 0.408      0.5        0.37037037 0.49019608
 0.4516129  0.5        0.39130435 0.36       0.6969697  0.6984127
 0.63636364 0.57692308 0.90740741 0.14285714 0.2173913  0.13888889
 0.7816092  0.78431373 0.15789474 0.59259259 0.7107438  0.66
 0.71538462 0.71428571 0.75       0.60655738 0.45588235 0.41052632
 0.24719101 0.68421053 0.67924528 0.         0.32       0.5
 0.19047619 0.25       0.42424242 0.53125    0.08333333 0.71428571
 0.63636364 0.22222222 0.64       0.16363636 0.13333333 0.48148148
 0.12962963 0.1969697  0.46478873 0.38461538 0.51612903 0.28571429
 0.47368421 0.56       0.76666667 0.48387097 0.47058824 0.35294118
 0.27272727 0.82926829 0.17142857 0.19047619 0.33333333 0.10526316
 0.51428571 0.36206897 0.59459459 0.63492063 0.22222222 0.16071429
 0.48648649 0.75609756 0.48837209 0.56481481 0.69158879 0.29032258
 0.07407407 0.07142857 0.53216374 0.51470588 0.34693878 0.60185185
 0.234375   0.6122449  0.52439024 0.68571429 0.52808989 0.68115942
 0.49122807 0.40659341 0.64705882 0.         0.67567568 0.42857143
 0.63207547 0.43333333 0.36363636 0.76595745 0.36585366 0.58333333
 0.79545455 0.74509804 0.38181818 0.475      0.36111111 0.57142857
 0.49056604 0.57407407 0.14482759 0.625      0.3        0.18518519
 0.28947368 0.26666667 0.67549669 0.4137931  0.14285714 0.41666667
 0.15254237 0.53947368 0.66       0.37209302 0.38636364 0.3559322
 0.58666667 0.42957746 0.67924528 0.17910448 0.484375   0.784
 0.64848485 0.27419355 0.38333333 0.75925926 0.52380952 0.19607843
 0.27868852 0.35714286 0.46534653 0.5        0.46296296 0.35
 0.32       0.4137931  0.65686275 0.72727273 0.91262136 0.86792453
 0.69444444 0.64864865 0.44761905 0.6618705  0.64963504 0.73684211]
upper bound: 0.48456300049059403
-----Evaluation is finished------
Class Accuracy 46.96%
Overall Prec@1 50.92% Prec@5 79.10%
/home/hs/anaconda3/envs/dzl/lib/python3.7/site-packages/numpy/core/_asarray.py:136: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray
  return array(a, dtype, copy=False, order=order, subok=True)

#20201019 PA with new conv & RGB

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
$ bash scripts/test/sthv1/Full.sh
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     RGB
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256

=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
video number:11522
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     PA
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256

=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
video number:11522
video 0 done, total 0/11522, average 0.242 sec/video, moving Prec@1 43.750 Prec@5 87.500
video 320 done, total 320/11522, average 0.034 sec/video, moving Prec@1 54.167 Prec@5 82.440
video 640 done, total 640/11522, average 0.029 sec/video, moving Prec@1 53.963 Prec@5 82.317
video 960 done, total 960/11522, average 0.027 sec/video, moving Prec@1 52.357 Prec@5 80.430
video 1280 done, total 1280/11522, average 0.026 sec/video, moving Prec@1 53.164 Prec@5 80.710
video 1600 done, total 1600/11522, average 0.026 sec/video, moving Prec@1 53.032 Prec@5 80.322
video 1920 done, total 1920/11522, average 0.025 sec/video, moving Prec@1 52.428 Prec@5 80.165
video 2240 done, total 2240/11522, average 0.025 sec/video, moving Prec@1 52.438 Prec@5 80.053
video 2560 done, total 2560/11522, average 0.025 sec/video, moving Prec@1 52.135 Prec@5 79.852
video 2880 done, total 2880/11522, average 0.025 sec/video, moving Prec@1 51.761 Prec@5 79.765
video 3200 done, total 3200/11522, average 0.025 sec/video, moving Prec@1 51.182 Prec@5 79.789
video 3520 done, total 3520/11522, average 0.025 sec/video, moving Prec@1 51.131 Prec@5 79.751
video 3840 done, total 3840/11522, average 0.025 sec/video, moving Prec@1 51.193 Prec@5 79.357
video 4160 done, total 4160/11522, average 0.024 sec/video, moving Prec@1 50.958 Prec@5 79.286
video 4480 done, total 4480/11522, average 0.024 sec/video, moving Prec@1 50.845 Prec@5 79.404
video 4800 done, total 4800/11522, average 0.024 sec/video, moving Prec@1 50.602 Prec@5 79.111
video 5120 done, total 5120/11522, average 0.024 sec/video, moving Prec@1 50.389 Prec@5 78.894
video 5440 done, total 5440/11522, average 0.024 sec/video, moving Prec@1 50.513 Prec@5 79.051
video 5760 done, total 5760/11522, average 0.024 sec/video, moving Prec@1 50.623 Prec@5 78.895
video 6080 done, total 6080/11522, average 0.024 sec/video, moving Prec@1 50.853 Prec@5 79.003
video 6400 done, total 6400/11522, average 0.024 sec/video, moving Prec@1 50.982 Prec@5 79.193
video 6720 done, total 6720/11522, average 0.024 sec/video, moving Prec@1 50.772 Prec@5 79.097
video 7040 done, total 7040/11522, average 0.024 sec/video, moving Prec@1 50.779 Prec@5 79.025
video 7360 done, total 7360/11522, average 0.024 sec/video, moving Prec@1 50.786 Prec@5 78.959
video 7680 done, total 7680/11522, average 0.024 sec/video, moving Prec@1 50.910 Prec@5 79.158
video 8000 done, total 8000/11522, average 0.024 sec/video, moving Prec@1 50.836 Prec@5 79.154
video 8320 done, total 8320/11522, average 0.024 sec/video, moving Prec@1 50.852 Prec@5 79.163
video 8640 done, total 8640/11522, average 0.024 sec/video, moving Prec@1 50.716 Prec@5 79.124
video 8960 done, total 8960/11522, average 0.024 sec/video, moving Prec@1 50.657 Prec@5 79.133
video 9280 done, total 9280/11522, average 0.024 sec/video, moving Prec@1 50.592 Prec@5 79.045
video 9600 done, total 9600/11522, average 0.024 sec/video, moving Prec@1 50.759 Prec@5 79.077
video 9920 done, total 9920/11522, average 0.024 sec/video, moving Prec@1 50.815 Prec@5 79.116
video 10240 done, total 10240/11522, average 0.024 sec/video, moving Prec@1 50.907 Prec@5 79.280
video 10560 done, total 10560/11522, average 0.024 sec/video, moving Prec@1 51.031 Prec@5 79.255
video 10880 done, total 10880/11522, average 0.024 sec/video, moving Prec@1 51.101 Prec@5 79.314
video 11200 done, total 11200/11522, average 0.024 sec/video, moving Prec@1 51.213 Prec@5 79.333
video 11520 done, total 11520/11522, average 0.024 sec/video, moving Prec@1 51.180 Prec@5 79.283
[0.91044776 0.3125     0.26666667 0.59701493 0.34210526 0.48275862
 0.73134328 0.5        0.63106796 0.70873786 0.44       0.48611111
 0.31914894 0.41935484 0.75159236 0.44       0.20481928 0.3125
 0.46341463 0.4        0.416      0.5        0.38888889 0.49019608
 0.4516129  0.41666667 0.39130435 0.36       0.6969697  0.68253968
 0.63636364 0.53846154 0.90740741 0.14285714 0.23913043 0.13888889
 0.8045977  0.80392157 0.13157895 0.59259259 0.7107438  0.64
 0.73076923 0.71428571 0.75       0.60655738 0.45588235 0.43157895
 0.26966292 0.70676692 0.6509434  0.         0.32       0.45
 0.19047619 0.25       0.45454545 0.53125    0.08333333 0.73469388
 0.6969697  0.22222222 0.64       0.18181818 0.13333333 0.5
 0.12962963 0.21212121 0.46478873 0.34615385 0.51612903 0.21428571
 0.47368421 0.56       0.83333333 0.4516129  0.29411765 0.41176471
 0.22727273 0.82926829 0.17142857 0.19047619 0.3        0.10526316
 0.54285714 0.34482759 0.59459459 0.63492063 0.16666667 0.16071429
 0.54054054 0.75609756 0.46511628 0.60185185 0.6635514  0.29032258
 0.07407407 0.07142857 0.52046784 0.51470588 0.32653061 0.58333333
 0.234375   0.59183673 0.56097561 0.6952381  0.57303371 0.68115942
 0.49122807 0.3956044  0.64705882 0.         0.67567568 0.38095238
 0.63207547 0.46666667 0.36363636 0.75531915 0.3902439  0.60416667
 0.78409091 0.76470588 0.39090909 0.475      0.36111111 0.53061224
 0.49056604 0.57407407 0.15172414 0.61538462 0.34       0.11111111
 0.28947368 0.26666667 0.66225166 0.43103448 0.14285714 0.4
 0.13559322 0.52631579 0.67       0.39534884 0.38636364 0.33050847
 0.57333333 0.42253521 0.68867925 0.17910448 0.53125    0.768
 0.64848485 0.29032258 0.38333333 0.77777778 0.51190476 0.19607843
 0.29508197 0.37142857 0.5049505  0.46875    0.44444444 0.325
 0.34       0.44827586 0.66666667 0.72727273 0.91262136 0.87735849
 0.72222222 0.64864865 0.43809524 0.64028777 0.65693431 0.73684211]
upper bound: 0.4842930969956842
-----Evaluation is finished------
Class Accuracy 46.93%
Overall Prec@1 51.18% Prec@5 79.28%
/home/hs/anaconda3/envs/dzl/lib/python3.7/site-packages/numpy/core/_asarray.py:136: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray
  return array(a, dtype, copy=False, order=order, subok=True)

#20201021 RGBo

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
$ python test_models.py something --VAP --batch_size=8 -j=4 --test_crops=1 --test_segments=8 --weights=pretrained/PAN_RGB_something_resnet50_shift8_blockres_avg_segment8_e50.pth.tar
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     RGB
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256

=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
video number:11522
video 0 done, total 0/11522, average 3.718 sec/video, moving Prec@1 25.000 Prec@5 62.500
video 160 done, total 160/11522, average 0.239 sec/video, moving Prec@1 46.429 Prec@5 76.786
video 320 done, total 320/11522, average 0.193 sec/video, moving Prec@1 47.866 Prec@5 76.220
video 480 done, total 480/11522, average 0.190 sec/video, moving Prec@1 47.951 Prec@5 78.074
video 640 done, total 640/11522, average 0.176 sec/video, moving Prec@1 46.759 Prec@5 76.698
video 800 done, total 800/11522, average 0.168 sec/video, moving Prec@1 45.421 Prec@5 75.248
video 960 done, total 960/11522, average 0.159 sec/video, moving Prec@1 45.971 Prec@5 74.690
video 1120 done, total 1120/11522, average 0.152 sec/video, moving Prec@1 46.099 Prec@5 74.734
video 1280 done, total 1280/11522, average 0.144 sec/video, moving Prec@1 46.429 Prec@5 75.233
video 1440 done, total 1440/11522, average 0.140 sec/video, moving Prec@1 46.685 Prec@5 74.862
video 1600 done, total 1600/11522, average 0.139 sec/video, moving Prec@1 46.766 Prec@5 74.876
video 1760 done, total 1760/11522, average 0.139 sec/video, moving Prec@1 45.871 Prec@5 74.321
video 1920 done, total 1920/11522, average 0.135 sec/video, moving Prec@1 45.747 Prec@5 74.274
video 2080 done, total 2080/11522, average 0.133 sec/video, moving Prec@1 45.546 Prec@5 74.234
video 2240 done, total 2240/11522, average 0.132 sec/video, moving Prec@1 45.418 Prec@5 74.199
video 2400 done, total 2400/11522, average 0.131 sec/video, moving Prec@1 45.307 Prec@5 74.211
video 2560 done, total 2560/11522, average 0.130 sec/video, moving Prec@1 45.132 Prec@5 74.182
video 2720 done, total 2720/11522, average 0.128 sec/video, moving Prec@1 45.345 Prec@5 74.450
video 2880 done, total 2880/11522, average 0.126 sec/video, moving Prec@1 44.910 Prec@5 74.100
video 3040 done, total 3040/11522, average 0.124 sec/video, moving Prec@1 44.980 Prec@5 74.180
video 3200 done, total 3200/11522, average 0.124 sec/video, moving Prec@1 44.514 Prec@5 74.034
video 3360 done, total 3360/11522, average 0.122 sec/video, moving Prec@1 44.626 Prec@5 74.228
video 3520 done, total 3520/11522, average 0.121 sec/video, moving Prec@1 44.586 Prec@5 74.093
video 3680 done, total 3680/11522, average 0.119 sec/video, moving Prec@1 44.306 Prec@5 73.861
video 3840 done, total 3840/11522, average 0.119 sec/video, moving Prec@1 44.491 Prec@5 73.831
video 4000 done, total 4000/11522, average 0.118 sec/video, moving Prec@1 44.212 Prec@5 73.802
video 4160 done, total 4160/11522, average 0.117 sec/video, moving Prec@1 44.218 Prec@5 73.800
video 4320 done, total 4320/11522, average 0.117 sec/video, moving Prec@1 44.362 Prec@5 73.706
video 4480 done, total 4480/11522, average 0.116 sec/video, moving Prec@1 44.385 Prec@5 73.708
video 4640 done, total 4640/11522, average 0.115 sec/video, moving Prec@1 44.535 Prec@5 73.580
video 4800 done, total 4800/11522, average 0.114 sec/video, moving Prec@1 44.468 Prec@5 73.440
video 4960 done, total 4960/11522, average 0.113 sec/video, moving Prec@1 44.304 Prec@5 73.289
video 5120 done, total 5120/11522, average 0.113 sec/video, moving Prec@1 44.423 Prec@5 73.362
video 5280 done, total 5280/11522, average 0.112 sec/video, moving Prec@1 44.402 Prec@5 73.298
video 5440 done, total 5440/11522, average 0.112 sec/video, moving Prec@1 44.475 Prec@5 73.477
video 5600 done, total 5600/11522, average 0.112 sec/video, moving Prec@1 44.454 Prec@5 73.520
video 5760 done, total 5760/11522, average 0.113 sec/video, moving Prec@1 44.435 Prec@5 73.370
video 5920 done, total 5920/11522, average 0.116 sec/video, moving Prec@1 44.518 Prec@5 73.482
video 6080 done, total 6080/11522, average 0.117 sec/video, moving Prec@1 44.514 Prec@5 73.489
video 6240 done, total 6240/11522, average 0.117 sec/video, moving Prec@1 44.542 Prec@5 73.464
video 6400 done, total 6400/11522, average 0.117 sec/video, moving Prec@1 44.538 Prec@5 73.533
video 6560 done, total 6560/11522, average 0.116 sec/video, moving Prec@1 44.488 Prec@5 73.538
video 6720 done, total 6720/11522, average 0.116 sec/video, moving Prec@1 44.471 Prec@5 73.484
video 6880 done, total 6880/11522, average 0.115 sec/video, moving Prec@1 44.440 Prec@5 73.432
video 7040 done, total 7040/11522, average 0.115 sec/video, moving Prec@1 44.509 Prec@5 73.411
video 7200 done, total 7200/11522, average 0.114 sec/video, moving Prec@1 44.478 Prec@5 73.432
video 7360 done, total 7360/11522, average 0.114 sec/video, moving Prec@1 44.435 Prec@5 73.358
video 7520 done, total 7520/11522, average 0.114 sec/video, moving Prec@1 44.474 Prec@5 73.433
video 7680 done, total 7680/11522, average 0.113 sec/video, moving Prec@1 44.511 Prec@5 73.452
video 7840 done, total 7840/11522, average 0.113 sec/video, moving Prec@1 44.521 Prec@5 73.471
video 8000 done, total 8000/11522, average 0.112 sec/video, moving Prec@1 44.456 Prec@5 73.489
video 8160 done, total 8160/11522, average 0.111 sec/video, moving Prec@1 44.478 Prec@5 73.470
video 8320 done, total 8320/11522, average 0.111 sec/video, moving Prec@1 44.549 Prec@5 73.487
video 8480 done, total 8480/11522, average 0.111 sec/video, moving Prec@1 44.475 Prec@5 73.445
video 8640 done, total 8640/11522, average 0.110 sec/video, moving Prec@1 44.438 Prec@5 73.462
video 8800 done, total 8800/11522, average 0.110 sec/video, moving Prec@1 44.550 Prec@5 73.490
video 8960 done, total 8960/11522, average 0.109 sec/video, moving Prec@1 44.625 Prec@5 73.550
video 9120 done, total 9120/11522, average 0.109 sec/video, moving Prec@1 44.665 Prec@5 73.510
video 9280 done, total 9280/11522, average 0.109 sec/video, moving Prec@1 44.660 Prec@5 73.482
video 9440 done, total 9440/11522, average 0.108 sec/video, moving Prec@1 44.634 Prec@5 73.455
video 9600 done, total 9600/11522, average 0.108 sec/video, moving Prec@1 44.775 Prec@5 73.564
video 9760 done, total 9760/11522, average 0.107 sec/video, moving Prec@1 44.799 Prec@5 73.618
video 9920 done, total 9920/11522, average 0.106 sec/video, moving Prec@1 44.722 Prec@5 73.580
video 10080 done, total 10080/11522, average 0.106 sec/video, moving Prec@1 44.826 Prec@5 73.672
video 10240 done, total 10240/11522, average 0.105 sec/video, moving Prec@1 44.838 Prec@5 73.800
video 10400 done, total 10400/11522, average 0.105 sec/video, moving Prec@1 44.860 Prec@5 73.741
video 10560 done, total 10560/11522, average 0.104 sec/video, moving Prec@1 44.909 Prec@5 73.723
video 10720 done, total 10720/11522, average 0.104 sec/video, moving Prec@1 44.976 Prec@5 73.723
video 10880 done, total 10880/11522, average 0.104 sec/video, moving Prec@1 44.994 Prec@5 73.778
video 11040 done, total 11040/11522, average 0.104 sec/video, moving Prec@1 44.904 Prec@5 73.724
video 11200 done, total 11200/11522, average 0.103 sec/video, moving Prec@1 44.995 Prec@5 73.742
video 11360 done, total 11360/11522, average 0.103 sec/video, moving Prec@1 44.986 Prec@5 73.716
video 11520 done, total 11520/11522, average 0.102 sec/video, moving Prec@1 44.975 Prec@5 73.685
[0.7761194  0.275      0.19047619 0.62686567 0.23684211 0.40517241
 0.62686567 0.44117647 0.45631068 0.6407767  0.37333333 0.38888889
 0.24468085 0.32258065 0.7133758  0.36       0.19879518 0.225
 0.39837398 0.28       0.32       0.48275862 0.40740741 0.43137255
 0.32258065 0.33333333 0.39130435 0.28       0.60606061 0.53968254
 0.63636364 0.57692308 0.72222222 0.10989011 0.23913043 0.16666667
 0.7816092  0.78431373 0.05263158 0.48148148 0.62809917 0.54
 0.66923077 0.63265306 0.59615385 0.52459016 0.41176471 0.34736842
 0.15730337 0.66165414 0.56603774 0.         0.36       0.4
 0.19047619 0.29166667 0.39393939 0.40625    0.         0.73469388
 0.60606061 0.13888889 0.6        0.11818182 0.1        0.38888889
 0.11111111 0.15151515 0.43661972 0.30769231 0.48387097 0.25
 0.5        0.44       0.7        0.41935484 0.29411765 0.29411765
 0.13636364 0.63414634 0.11428571 0.14285714 0.33333333 0.10526316
 0.62857143 0.29310345 0.52702703 0.55555556 0.16666667 0.10714286
 0.48648649 0.68292683 0.41860465 0.46296296 0.57009346 0.27419355
 0.03703704 0.11904762 0.41520468 0.41176471 0.24489796 0.51851852
 0.21875    0.46938776 0.42682927 0.6        0.5505618  0.75362319
 0.47368421 0.2967033  0.54901961 0.11764706 0.60810811 0.38095238
 0.66037736 0.36666667 0.21818182 0.62765957 0.2195122  0.58333333
 0.79545455 0.54901961 0.35454545 0.425      0.36111111 0.44897959
 0.49056604 0.53703704 0.15862069 0.53846154 0.22       0.18518519
 0.13157895 0.23333333 0.64238411 0.29310345 0.07142857 0.45
 0.08474576 0.51315789 0.57       0.34883721 0.25       0.38135593
 0.56       0.33098592 0.6509434  0.17910448 0.484375   0.752
 0.56363636 0.12903226 0.28333333 0.7037037  0.5        0.15686275
 0.18032787 0.21428571 0.43564356 0.4375     0.24074074 0.25
 0.26       0.31034483 0.60784314 0.66666667 0.83495146 0.75471698
 0.68518519 0.45945946 0.44761905 0.61870504 0.59124088 0.71052632]
upper bound: 0.42524204990350917
-----Evaluation is finished------
Class Accuracy 40.94%
Overall Prec@1 44.97% Prec@5 73.69%
E:\Program Files\Anaconda3\envs\tsm\lib\site-packages\numpy\core\_asarray.py:136: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray  return array(a, dtype, copy=False, order=order, subok=True)

#20201021 RGBo & PAn & Liten

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
# python test_models.py something --VAP --batch_size=16 -j=4 --test_crops=1 --test_segments=8,8,8 --weights=
# pretrained/PAN_Lite_something_resnet50_shift8_blockres_avg_segment8_e80.pth.tar,
# pretrained/PAN_RGB_something_resnet50_shift8_blockres_avg_segment8_e50.pth.tar,
# pretrained/PAN_PA_something_resnet50_shift8_blockres_avg_segment8_e80.pth.tar --full_res --twice_sample
$ bash scripts/test/sthv1/En.sh
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     Lite
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256

=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
=> Converting the ImageNet model to a PAN_Lite init model
=> Done. PAN_lite model ready...
=> Using twice sample for the dataset...
video number:11522
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     RGB
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256

=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
=> Using twice sample for the dataset...
video number:11522
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     PA
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256

=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
=> Using twice sample for the dataset...
video number:11522
video 0 done, total 0/11522, average 1.035 sec/video, moving Prec@1 50.000 Prec@5 81.250
video 320 done, total 320/11522, average 0.156 sec/video, moving Prec@1 54.464 Prec@5 86.012
video 640 done, total 640/11522, average 0.151 sec/video, moving Prec@1 56.707 Prec@5 84.299
video 960 done, total 960/11522, average 0.150 sec/video, moving Prec@1 55.635 Prec@5 82.377
video 1280 done, total 1280/11522, average 0.148 sec/video, moving Prec@1 56.481 Prec@5 83.102
video 1600 done, total 1600/11522, average 0.148 sec/video, moving Prec@1 56.621 Prec@5 83.106
video 1920 done, total 1920/11522, average 0.146 sec/video, moving Prec@1 55.682 Prec@5 82.800
video 2240 done, total 2240/11522, average 0.146 sec/video, moving Prec@1 55.629 Prec@5 82.624
video 2560 done, total 2560/11522, average 0.145 sec/video, moving Prec@1 55.668 Prec@5 82.026
video 2880 done, total 2880/11522, average 0.145 sec/video, moving Prec@1 55.249 Prec@5 82.113
video 3200 done, total 3200/11522, average 0.145 sec/video, moving Prec@1 54.851 Prec@5 82.121
video 3520 done, total 3520/11522, average 0.145 sec/video, moving Prec@1 54.666 Prec@5 82.070
video 3840 done, total 3840/11522, average 0.144 sec/video, moving Prec@1 54.876 Prec@5 81.743
video 4160 done, total 4160/11522, average 0.144 sec/video, moving Prec@1 54.741 Prec@5 81.561
video 4480 done, total 4480/11522, average 0.144 sec/video, moving Prec@1 54.560 Prec@5 81.539
video 4800 done, total 4800/11522, average 0.143 sec/video, moving Prec@1 54.360 Prec@5 81.333
video 5120 done, total 5120/11522, average 0.143 sec/video, moving Prec@1 53.972 Prec@5 81.055
video 5440 done, total 5440/11522, average 0.143 sec/video, moving Prec@1 54.271 Prec@5 81.305
video 5760 done, total 5760/11522, average 0.143 sec/video, moving Prec@1 54.276 Prec@5 81.198
video 6080 done, total 6080/11522, average 0.143 sec/video, moving Prec@1 54.068 Prec@5 81.283
video 6400 done, total 6400/11522, average 0.143 sec/video, moving Prec@1 54.193 Prec@5 81.390
video 6720 done, total 6720/11522, average 0.143 sec/video, moving Prec@1 54.097 Prec@5 81.309
video 7040 done, total 7040/11522, average 0.143 sec/video, moving Prec@1 53.997 Prec@5 81.278
video 7360 done, total 7360/11522, average 0.143 sec/video, moving Prec@1 53.932 Prec@5 81.223
video 7680 done, total 7680/11522, average 0.143 sec/video, moving Prec@1 53.963 Prec@5 81.302
video 8000 done, total 8000/11522, average 0.143 sec/video, moving Prec@1 53.905 Prec@5 81.275
video 8320 done, total 8320/11522, average 0.143 sec/video, moving Prec@1 53.839 Prec@5 81.310
video 8640 done, total 8640/11522, average 0.143 sec/video, moving Prec@1 53.662 Prec@5 81.181
video 8960 done, total 8960/11522, average 0.143 sec/video, moving Prec@1 53.632 Prec@5 81.205
video 9280 done, total 9280/11522, average 0.143 sec/video, moving Prec@1 53.647 Prec@5 81.185
video 9600 done, total 9600/11522, average 0.143 sec/video, moving Prec@1 53.744 Prec@5 81.198
video 9920 done, total 9920/11522, average 0.143 sec/video, moving Prec@1 53.744 Prec@5 81.180
video 10240 done, total 10240/11522, average 0.143 sec/video, moving Prec@1 53.890 Prec@5 81.318
video 10560 done, total 10560/11522, average 0.142 sec/video, moving Prec@1 53.943 Prec@5 81.288
video 10880 done, total 10880/11522, average 0.142 sec/video, moving Prec@1 53.992 Prec@5 81.360
video 11200 done, total 11200/11522, average 0.142 sec/video, moving Prec@1 54.012 Prec@5 81.419
video 11520 done, total 11520/11522, average 0.142 sec/video, moving Prec@1 54.044 Prec@5 81.427
[0.89552239 0.3875     0.36190476 0.65671642 0.44736842 0.50862069
 0.69402985 0.47058824 0.74757282 0.76699029 0.44       0.63888889
 0.41489362 0.38709677 0.81528662 0.56       0.21084337 0.3125
 0.5203252  0.4        0.416      0.67241379 0.40740741 0.43137255
 0.51612903 0.41666667 0.47826087 0.44       0.72727273 0.6984127
 0.68181818 0.61538462 0.85185185 0.23076923 0.32608696 0.16666667
 0.82758621 0.82352941 0.13157895 0.7037037  0.70247934 0.68
 0.71538462 0.67346939 0.73076923 0.63934426 0.46323529 0.47368421
 0.30337079 0.65413534 0.6509434  0.         0.32       0.55
 0.14285714 0.29166667 0.48484848 0.59375    0.         0.75510204
 0.75757576 0.25       0.64       0.15454545 0.16666667 0.58333333
 0.11111111 0.21212121 0.45070423 0.38461538 0.5483871  0.28571429
 0.5        0.6        0.85       0.58064516 0.23529412 0.41176471
 0.36363636 0.80487805 0.14285714 0.0952381  0.4        0.10526316
 0.62857143 0.39655172 0.64864865 0.80952381 0.22222222 0.10714286
 0.56756757 0.7804878  0.60465116 0.65740741 0.75700935 0.30645161
 0.07407407 0.19047619 0.57894737 0.57352941 0.37755102 0.58333333
 0.25       0.69387755 0.69512195 0.72380952 0.59550562 0.72463768
 0.38596491 0.3956044  0.60784314 0.11764706 0.67567568 0.47619048
 0.66037736 0.53333333 0.29090909 0.74468085 0.36585366 0.64583333
 0.81818182 0.60784314 0.43636364 0.475      0.36111111 0.55102041
 0.43396226 0.59259259 0.13793103 0.65384615 0.3        0.25925926
 0.34210526 0.23333333 0.70860927 0.5        0.14285714 0.31666667
 0.18644068 0.60526316 0.69       0.34883721 0.54545455 0.47457627
 0.50666667 0.48591549 0.61320755 0.20895522 0.53125    0.76
 0.67878788 0.24193548 0.43333333 0.90740741 0.60714286 0.2745098
 0.26229508 0.41428571 0.48514851 0.34375    0.48148148 0.425
 0.34       0.48275862 0.73529412 0.81818182 0.9223301  0.9245283
 0.72222222 0.72972973 0.44761905 0.6618705  0.73722628 0.69736842]
upper bound: 0.5131311156873893
-----Evaluation is finished------
Class Accuracy 49.83%
Overall Prec@1 54.04% Prec@5 81.43%
/home/hs/anaconda3/envs/dzl/lib/python3.7/site-packages/numpy/core/_asarray.py:136: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray
  return array(a, dtype, copy=False, order=order, subok=True)

#20201021 PA with conv & RGBo & Liten

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
# python test_models.py something --VAP --batch_size=16 -j=4 --test_crops=1 --test_segments=8,8,8 --weights=
# pretrained/PAN_Lite_something_resnet50_shift8_blockres_avg_segment8_e80.pth.tar,
# pretrained/PAN_RGB_something_resnet50_shift8_blockres_avg_segment8_e50.pth.tar,
# temp/PAN_APA_something_resnet50_shift8_blockres_avg_segment8_e80.pth.tar --full_res --twice_sample
$ bash scripts/test/sthv1/En.sh 
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     Lite
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256

=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
=> Converting the ImageNet model to a PAN_Lite init model
=> Done. PAN_lite model ready...
=> Using twice sample for the dataset...
video number:11522
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     RGB
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256

=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
=> Using twice sample for the dataset...
video number:11522
something: 174 classes
=> shift: True, shift_div: 8, shift_place: blockres

    Initializing PAN with base model: resnet50.
    PAN Configurations:
        input_modality:     PA
        num_segments:       8
        new_length:         1
        consensus_module:   avg
        dropout_ratio:      0.8
        img_feature_dim:    256

=> base model: resnet50
Adding temporal shift...
=> Using 3-level VAP
=> Using twice sample for the dataset...
video number:11522
video 0 done, total 0/11522, average 1.103 sec/video, moving Prec@1 50.000 Prec@5 81.250
video 320 done, total 320/11522, average 0.156 sec/video, moving Prec@1 55.655 Prec@5 84.226
video 640 done, total 640/11522, average 0.153 sec/video, moving Prec@1 56.707 Prec@5 83.232
video 960 done, total 960/11522, average 0.152 sec/video, moving Prec@1 54.713 Prec@5 82.070
video 1280 done, total 1280/11522, average 0.152 sec/video, moving Prec@1 55.247 Prec@5 82.793
video 1600 done, total 1600/11522, average 0.150 sec/video, moving Prec@1 55.384 Prec@5 82.859
video 1920 done, total 1920/11522, average 0.151 sec/video, moving Prec@1 54.700 Prec@5 82.593
video 2240 done, total 2240/11522, average 0.150 sec/video, moving Prec@1 55.363 Prec@5 82.314
video 2560 done, total 2560/11522, average 0.151 sec/video, moving Prec@1 55.124 Prec@5 81.910
video 2880 done, total 2880/11522, average 0.150 sec/video, moving Prec@1 54.696 Prec@5 81.975
video 3200 done, total 3200/11522, average 0.149 sec/video, moving Prec@1 54.353 Prec@5 81.996
video 3520 done, total 3520/11522, average 0.150 sec/video, moving Prec@1 54.157 Prec@5 81.957
video 3840 done, total 3840/11522, average 0.149 sec/video, moving Prec@1 54.253 Prec@5 81.613
video 4160 done, total 4160/11522, average 0.149 sec/video, moving Prec@1 54.191 Prec@5 81.513
video 4480 done, total 4480/11522, average 0.148 sec/video, moving Prec@1 54.070 Prec@5 81.606
video 4800 done, total 4800/11522, average 0.148 sec/video, moving Prec@1 53.862 Prec@5 81.499
video 5120 done, total 5120/11522, average 0.148 sec/video, moving Prec@1 53.563 Prec@5 81.211
video 5440 done, total 5440/11522, average 0.148 sec/video, moving Prec@1 53.757 Prec@5 81.378
video 5760 done, total 5760/11522, average 0.148 sec/video, moving Prec@1 53.826 Prec@5 81.302
video 6080 done, total 6080/11522, average 0.148 sec/video, moving Prec@1 53.724 Prec@5 81.365
video 6400 done, total 6400/11522, average 0.147 sec/video, moving Prec@1 53.819 Prec@5 81.468
video 6720 done, total 6720/11522, average 0.147 sec/video, moving Prec@1 53.593 Prec@5 81.517
video 7040 done, total 7040/11522, average 0.147 sec/video, moving Prec@1 53.543 Prec@5 81.519
video 7360 done, total 7360/11522, average 0.147 sec/video, moving Prec@1 53.484 Prec@5 81.386
video 7680 done, total 7680/11522, average 0.147 sec/video, moving Prec@1 53.495 Prec@5 81.510
video 8000 done, total 8000/11522, average 0.147 sec/video, moving Prec@1 53.406 Prec@5 81.562
video 8320 done, total 8320/11522, average 0.147 sec/video, moving Prec@1 53.491 Prec@5 81.670
video 8640 done, total 8640/11522, average 0.147 sec/video, moving Prec@1 53.373 Prec@5 81.573
video 8960 done, total 8960/11522, average 0.147 sec/video, moving Prec@1 53.376 Prec@5 81.584
video 9280 done, total 9280/11522, average 0.146 sec/video, moving Prec@1 53.292 Prec@5 81.616
video 9600 done, total 9600/11522, average 0.146 sec/video, moving Prec@1 53.380 Prec@5 81.614
video 9920 done, total 9920/11522, average 0.146 sec/video, moving Prec@1 53.422 Prec@5 81.542
video 10240 done, total 10240/11522, average 0.146 sec/video, moving Prec@1 53.539 Prec@5 81.699
video 10560 done, total 10560/11522, average 0.146 sec/video, moving Prec@1 53.669 Prec@5 81.647
video 10880 done, total 10880/11522, average 0.146 sec/video, moving Prec@1 53.744 Prec@5 81.681
video 11200 done, total 11200/11522, average 0.146 sec/video, moving Prec@1 53.736 Prec@5 81.705
video 11520 done, total 11520/11522, average 0.145 sec/video, moving Prec@1 53.810 Prec@5 81.739
[0.92537313 0.4        0.31428571 0.68656716 0.39473684 0.52586207
 0.70895522 0.52941176 0.73786408 0.75728155 0.49333333 0.56944444
 0.45744681 0.35483871 0.79617834 0.46       0.22289157 0.3375
 0.4796748  0.4        0.432      0.55172414 0.40740741 0.43137255
 0.51612903 0.5        0.52173913 0.24       0.72727273 0.71428571
 0.68181818 0.57692308 0.88888889 0.18681319 0.26086957 0.13888889
 0.8045977  0.83333333 0.13157895 0.66666667 0.73553719 0.66
 0.71538462 0.75510204 0.71153846 0.67213115 0.54411765 0.50526316
 0.30337079 0.71428571 0.64150943 0.         0.36       0.55
 0.23809524 0.33333333 0.36363636 0.5        0.08333333 0.75510204
 0.72727273 0.27777778 0.62       0.11818182 0.16666667 0.55555556
 0.11111111 0.1969697  0.47887324 0.42307692 0.5483871  0.25
 0.5        0.64       0.85       0.61290323 0.29411765 0.41176471
 0.36363636 0.85365854 0.14285714 0.11904762 0.36666667 0.15789474
 0.6        0.39655172 0.68918919 0.73015873 0.11111111 0.10714286
 0.51351351 0.75609756 0.55813953 0.66666667 0.75700935 0.22580645
 0.07407407 0.16666667 0.57309942 0.55882353 0.35714286 0.58333333
 0.28125    0.59183673 0.67073171 0.73333333 0.61797753 0.73913043
 0.45614035 0.40659341 0.62745098 0.         0.64864865 0.42857143
 0.64150943 0.5        0.29090909 0.77659574 0.41463415 0.625
 0.79545455 0.68627451 0.41818182 0.5        0.25       0.59183673
 0.43396226 0.64814815 0.14482759 0.68269231 0.32       0.22222222
 0.36842105 0.23333333 0.68874172 0.51724138 0.21428571 0.3
 0.18644068 0.56578947 0.68       0.39534884 0.54545455 0.44915254
 0.54666667 0.50704225 0.66037736 0.19402985 0.5        0.76
 0.69090909 0.24193548 0.35       0.88888889 0.60714286 0.21568627
 0.26229508 0.37142857 0.43564356 0.4375     0.44444444 0.35
 0.36       0.51724138 0.67647059 0.81818182 0.91262136 0.9245283
 0.74074074 0.7027027  0.46666667 0.65467626 0.71532847 0.71052632]
upper bound: 0.5095849479276692
-----Evaluation is finished------
Class Accuracy 49.46%
Overall Prec@1 53.81% Prec@5 81.74%
/home/hs/anaconda3/envs/dzl/lib/python3.7/site-packages/numpy/core/_asarray.py:136: VisibleDeprecationWarning: Creating an ndarray from ragged nested sequences (which is a list-or-tuple of lists-or-tuples-or ndarrays with different lengths or shapes) is deprecated. If you meant to do this, you must specify 'dtype=object' when creating the ndarray
  return array(a, dtype, copy=False, order=order, subok=True)