模型库统计¶
在本页面中，我们列举了我们支持的所有算法。你可以点击链接跳转至对应的模型详情页面。
另外，我们还列出了我们提供的所有模型权重文件。你可以使用排序和搜索功能找到需要的模型权重，并使用链接跳转至模型详情页面。
所有已支持的算法¶
论文数量：35
Algorithm: 35
模型权重文件数量：189
[Algorithm] Actor-Centric Relation Network (2 ckpts)
[Algorithm] Long-Term Feature Banks for Detailed Video Understanding (2 ckpts)
[Algorithm] SlowFast Networks for Video Recognition (7 ckpts)
[Algorithm] SlowFast Networks for Video Recognition (15 ckpts)
[Algorithm] VideoMAE: Masked Autoencoders are Data-Efficient Learners for Self-Supervised Video Pre-Training (2 ckpts)
[Algorithm] Non-local Neural Networks (4 ckpts)
[Algorithm] Learning Spatiotemporal Features with 3D Convolutional Networks (1 ckpts)
[Algorithm] Video Classification with Channel-Separated Convolutional Networks (3 ckpts)
[Algorithm] Quo Vadis, Action Recognition? A New Model and the Kinetics Dataset (6 ckpts)
[Algorithm] MViTv2: Improved Multiscale Vision Transformers for Classification and Detection (9 ckpts)
[Algorithm] Omni-sourced Webly-supervised Learning for Video Recognition (1 ckpts)
[Algorithm] A Closer Look at Spatiotemporal Convolutions for Action Recognition (2 ckpts)
[Algorithm] SlowFast Networks for Video Recognition (5 ckpts)
[Algorithm] SlowFast Networks for Video Recognition (10 ckpts)
[Algorithm] Video Swin Transformer (6 ckpts)
[Algorithm] TAM: Temporal Adaptive Module for Video Recognition (3 ckpts)
[Algorithm] Is Space-Time Attention All You Need for Video Understanding (3 ckpts)
[Algorithm] Temporal Interlacing Network (3 ckpts)
[Algorithm] Temporal Pyramid Network for Action Recognition (3 ckpts)
[Algorithm] Temporal Relational Reasoning in Videos (2 ckpts)
[Algorithm] TSM: Temporal Shift Module for Efficient Video Understanding (12 ckpts)
[Algorithm] Temporal Segment Networks: Towards Good Practices for Deep Action Recognition (12 ckpts)
[Algorithm] UniFormer: Unified Transformer for Efficient Spatiotemporal Representation Learning (3 ckpts)
[Algorithm] UniFormerV2: Spatiotemporal Learning by Arming Image ViTs with Video UniFormer (23 ckpts)
[Algorithm] VideoMAE: Masked Autoencoders are Data-Efficient Learners for Self-Supervised Video Pre-Training (2 ckpts)
[Algorithm] VideoMAE V2: Scaling Video Masked Autoencoders with Dual Masking (2 ckpts)
[Algorithm] X3D: Expanding Architectures for Efficient Video Recognition (2 ckpts)
[Algorithm] Audiovisual SlowFast Networks for Video Recognition (1 ckpts)
[Algorithm] BMN: Boundary-Matching Network for Temporal Action Proposal Generation (2 ckpts)
[Algorithm] BSN: Boundary Sensitive Network for Temporal Action Proposal Generation (1 ckpts)
[Algorithm] CLIP4Clip: An Empirical Study of CLIP for End to End Video Clip Retrieval (1 ckpts)
[Algorithm] Two-Stream Adaptive Graph Convolutional Networks for Skeleton-Based Action Recognition (8 ckpts)
[Algorithm] Revisiting Skeleton-based Action Recognition (7 ckpts)
[Algorithm] Spatial Temporal Graph Convolutional Networks for Skeleton-Based Action Recognition (16 ckpts)
[Algorithm] PYSKL: Towards Good Practices for Skeleton Action Recognition (8 ckpts)
行为识别¶
Kinetics-400¶
模型 |
参数量 (M) |
Flops (G) |
Top-1 (%) |
Top-5 (%) |
Readme |
|---|---|---|---|---|---|
c2d_r50-in1k-pre-nopool_8xb32-8x8x1-100e_kinetics400-rgb |
24.30 |
33.00 |
73.44 |
91.0 |
|
c2d_r101-in1k-pre-nopool_8xb32-8x8x1-100e_kinetics400-rgb |
43.30 |
63.00 |
74.97 |
91.77 |
|
c2d_r50-in1k-pre_8xb32-8x8x1-100e_kinetics400-rgb |
24.30 |
19.00 |
73.89 |
91.21 |
|
c2d_r50-in1k-pre_8xb32-16x4x1-100e_kinetics400-rgb |
24.30 |
39.00 |
74.97 |
91.91 |
|
ircsn_ig65m-pretrained-r152_8xb12-32x2x1-58e_kinetics400-rgb |
29.70 |
97.63 |
82.87 |
95.9 |
|
ircsn_ig65m-pretrained-r152-bnfrozen_8xb12-32x2x1-58e_kinetics400-rgb |
29.70 |
97.63 |
82.84 |
95.92 |
|
ircsn_ig65m-pretrained-r50-bnfrozen_8xb12-32x2x1-58e_kinetics400-rgb |
13.13 |
55.90 |
79.44 |
94.26 |
|
ipcsn_r152_32x2x1-180e_kinetics400-rgb |
33.02 |
109.90 |
77.80 |
93.1 |
|
ircsn_r152_32x2x1-180e_kinetics400-rgb |
29.70 |
97.63 |
76.53 |
92.28 |
|
ipcsn_ig65m-pretrained-r152-bnfrozen_32x2x1-58e_kinetics400-rgb |
33.02 |
109.90 |
82.68 |
95.69 |
|
ipcsn_sports1m-pretrained-r152-bnfrozen_32x2x1-58e_kinetics400-rgb |
33.02 |
109.90 |
79.07 |
93.82 |
|
ircsn_sports1m-pretrained-r152-bnfrozen_32x2x1-58e_kinetics400-rgb |
33.02 |
109.90 |
78.57 |
93.44 |
|
i3d_imagenet-pretrained-r50-nl-dot-product_8xb8-32x2x1-100e_kinetics400-rgb |
35.40 |
59.30 |
74.80 |
92.07 |
|
i3d_imagenet-pretrained-r50-nl-embedded-gaussian_8xb8-32x2x1-100e_kinetics400-rgb |
35.40 |
59.30 |
74.73 |
91.8 |
|
i3d_imagenet-pretrained-r50-nl-gaussian_8xb8-32x2x1-100e_kinetics400-rgb |
31.70 |
56.50 |
73.97 |
91.33 |
|
i3d_imagenet-pretrained-r50_8xb8-32x2x1-100e_kinetics400-rgb |
28.00 |
43.50 |
73.47 |
91.27 |
|
i3d_imagenet-pretrained-r50_8xb8-dense-32x2x1-100e_kinetics400-rgb |
28.00 |
43.50 |
73.77 |
91.35 |
|
i3d_imagenet-pretrained-r50-heavy_8xb8-32x2x1-100e_kinetics400-rgb |
33.00 |
166.30 |
76.21 |
92.48 |
|
mvit-small-p244_32xb16-16x4x1-200e_kinetics400-rgb_infer |
81.10 |
94.7 |
|||
mvit-small-p244_32xb16-16x4x1-200e_kinetics400-rgb |
34.50 |
64.00 |
80.60 |
94.7 |
|
mvit-base-p244_32x3x1_kinetics400-rgb |
81.10 |
94.7 |
|||
mvit-large-p244_40x3x1_kinetics400-rgb |
81.10 |
94.7 |
|||
mvit-small-p244_k400-maskfeat-pre_8xb32-16x4x1-100e_kinetics400-rgb |
36.40 |
71.00 |
81.80 |
95.2 |
|
slowonly_r50_8xb16-8x8x1-256e_imagenet-kinetics400-rgb |
32.45 |
54.75 |
77.30 |
93.23 |
|
r2plus1d_r34_8xb8-8x8x1-180e_kinetics400-rgb |
63.80 |
53.10 |
69.76 |
88.41 |
|
r2plus1d_r34_8xb8-32x2x1-180e_kinetics400-rgb |
63.80 |
213.00 |
75.46 |
92.28 |
|
slowfast_r50_8xb8-4x16x1-256e_kinetics400-rgb |
34.50 |
36.30 |
75.55 |
92.35 |
|
slowfast_r50_8xb8-8x8x1-256e_kinetics400-rgb |
34.60 |
66.10 |
76.80 |
92.99 |
|
slowfast_r50_8xb8-8x8x1-steplr-256e_kinetics400-rgb |
34.60 |
66.10 |
76.65 |
92.86 |
|
slowfast_r101_8xb8-8x8x1-256e_kinetics400-rgb |
62.90 |
126.00 |
78.65 |
93.88 |
|
slowfast_r101-r50_32xb8-4x16x1-256e_kinetics400-rgb |
62.40 |
64.90 |
77.03 |
92.99 |
|
slowonly_r50_8xb16-4x16x1-256e_kinetics400-rgb |
32.45 |
27.38 |
72.68 |
90.68 |
|
slowonly_r50_8xb16-8x8x1-256e_kinetics400-rgb |
32.45 |
54.75 |
74.82 |
91.8 |
|
slowonly_r101_8xb16-8x8x1-196e_kinetics400-rgb |
60.36 |
112.00 |
76.28 |
92.7 |
|
slowonly_imagenet-pretrained-r50_8xb16-4x16x1-steplr-150e_kinetics400-rgb |
32.45 |
27.38 |
74.83 |
91.6 |
|
slowonly_imagenet-pretrained-r50_8xb16-8x8x1-steplr-150e_kinetics400-rgb |
32.45 |
54.75 |
75.96 |
92.4 |
|
slowonly_r50-in1k-pre-nl-embedded-gaussian_8xb16-4x16x1-steplr-150e_kinetics400-rgb |
39.81 |
43.23 |
74.84 |
91.41 |
|
slowonly_r50-in1k-pre-nl-embedded-gaussian_8xb16-8x8x1-steplr-150e_kinetics400-rgb |
39.81 |
96.66 |
76.35 |
92.18 |
|
swin-tiny-p244-w877_in1k-pre_8xb8-amp-32x2x1-30e_kinetics400-rgb |
28.20 |
88.00 |
78.90 |
93.77 |
|
swin-small-p244-w877_in1k-pre_8xb8-amp-32x2x1-30e_kinetics400-rgb |
49.80 |
166.00 |
80.54 |
94.46 |
|
swin-base-p244-w877_in1k-pre_8xb8-amp-32x2x1-30e_kinetics400-rgb |
88.00 |
282.00 |
80.57 |
94.49 |
|
swin-large-p244-w877_in22k-pre_8xb8-amp-32x2x1-30e_kinetics400-rgb |
197.00 |
604.00 |
83.46 |
95.91 |
|
tanet_imagenet-pretrained-r50_8xb8-dense-1x1x8-100e_kinetics400-rgb |
25.60 |
43.00 |
76.25 |
92.41 |
|
timesformer_divST_8xb8-8x32x1-15e_kinetics400-rgb |
196.00 |
77.69 |
93.45 |
||
timesformer_jointST_8xb8-8x32x1-15e_kinetics400-rgb |
180.00 |
76.95 |
93.28 |
||
timesformer_spaceOnly_8xb8-8x32x1-15e_kinetics400-rgb |
141.00 |
76.93 |
92.88 |
||
tin_kinetics400-pretrained-tsm-r50_1x1x8-50e_kinetics400-rgb |
24.36 |
32.97 |
71.86 |
90.44 |
|
tpn-slowonly_r50_8xb8-8x8x1-150e_kinetics400-rgb |
91.50 |
66.01 |
74.20 |
91.48 |
|
tpn-slowonly_imagenet-pretrained-r50_8xb8-8x8x1-150e_kinetics400-rgb |
91.50 |
66.01 |
76.74 |
92.57 |
|
tsm_imagenet-pretrained-r50_8xb16-1x1x8-50e_kinetics400-rgb |
23.87 |
32.88 |
73.18 |
90.56 |
|
tsm_imagenet-pretrained-r50_8xb16-1x1x8-100e_kinetics400-rgb |
23.87 |
32.88 |
73.22 |
90.22 |
|
tsm_imagenet-pretrained-r50_8xb16-1x1x16-50e_kinetics400-rgb |
23.87 |
65.75 |
75.12 |
91.55 |
|
tsm_imagenet-pretrained-r50_8xb16-dense-1x1x8-50e_kinetics400-rgb |
23.87 |
32.88 |
73.38 |
90.78 |
|
tsm_imagenet-pretrained-r50-nl-embedded-gaussian_8xb16-1x1x8-50e_kinetics400-rgb |
31.68 |
61.30 |
74.34 |
91.23 |
|
tsm_imagenet-pretrained-r50-nl-dot-product_8xb16-1x1x8-50e_kinetics400-rgb |
31.68 |
61.30 |
74.49 |
91.15 |
|
tsm_imagenet-pretrained-r50-nl-gaussian_8xb16-1x1x8-50e_kinetics400-rgb |
28.00 |
59.06 |
73.66 |
90.99 |
|
tsm_imagenet-pretrained-mobileone-s4_8xb16-1x1x16-50e_kinetics400-rgb |
13.72 |
48.65 |
74.38 |
91.71 |
|
tsm_imagenet-pretrained-r101_8xb16-1x1x8-50e_sthv2-rgb |
2.74 |
3.27 |
63.70 |
88.28 |
|
tsn_imagenet-pretrained-r50_8xb32-1x1x3-100e_kinetics400-rgb |
24.33 |
102.70 |
72.83 |
90.65 |
|
tsn_imagenet-pretrained-r50_8xb32-1x1x5-100e_kinetics400-rgb |
24.33 |
102.70 |
73.80 |
91.21 |
|
tsn_imagenet-pretrained-r50_8xb32-1x1x8-100e_kinetics400-rgb |
24.33 |
102.70 |
74.12 |
91.34 |
|
tsn_imagenet-pretrained-r50_8xb32-dense-1x1x5-100e_kinetics400-rgb |
24.33 |
102.70 |
71.37 |
89.67 |
|
tsn_imagenet-pretrained-r101_8xb32-1x1x8-100e_kinetics400-rgb |
43.32 |
195.80 |
75.89 |
92.07 |
|
tsn_imagenet-pretrained-rn101-32x4d_8xb32-1x1x3-100e_kinetics400-rgb |
42.95 |
200.30 |
72.95 |
90.36 |
|
tsn_imagenet-pretrained-dense161_8xb32-1x1x3-100e_kinetics400-rgb |
27.36 |
194.60 |
72.07 |
90.15 |
|
tsn_imagenet-pretrained-swin-transformer_8xb32-1x1x3-100e_kinetics400-rgb |
87.15 |
386.70 |
77.03 |
92.61 |
|
tsn_imagenet-pretrained-swin-transformer_32xb8-1x1x8-50e_kinetics400-rgb |
87.15 |
386.70 |
79.22 |
94.2 |
|
tsn_imagenet-pretrained-mobileone-s4_8xb32-1x1x8-100e_kinetics400-rgb |
13.72 |
76.00 |
73.65 |
91.32 |
|
tsn_imagenet-pretrained-r50_8xb32-1x1x8-50e_sthv2-rgb |
23.87 |
102.70 |
35.51 |
67.09 |
|
tsn_imagenet-pretrained-r50_8xb32-1x1x16-50e_sthv2-rgb |
23.87 |
102.70 |
36.91 |
68.77 |
|
uniformer-small_imagenet1k-pre_16x4x1_kinetics400-rgb |
80.90 |
94.6 |
|||
uniformer-base_imagenet1k-pre_16x4x1_kinetics400-rgb |
82.00 |
95.0 |
|||
uniformer-base_imagenet1k-pre_32x4x1_kinetics400-rgb |
83.10 |
95.3 |
|||
uniformerv2-base-p16-res224_clip_8xb32-u8_kinetics400-rgb |
84.30 |
96.4 |
|||
uniformerv2-base-p16-res224_clip-kinetics710-pre_8xb32-u8_kinetics400-rgb |
85.80 |
97.1 |
|||
uniformerv2-large-p14-res224_clip-kinetics710-pre_u8_kinetics400-rgb |
88.70 |
98.1 |
|||
uniformerv2-large-p14-res224_clip-kinetics710-pre_u16_kinetics400-rgb |
89.00 |
98.2 |
|||
uniformerv2-large-p14-res224_clip-kinetics710-pre_u32_kinetics400-rgb |
89.30 |
98.2 |
|||
uniformerv2-large-p14-res336_clip-kinetics710-pre_u32_kinetics400-rgb |
89.50 |
98.4 |
|||
uniformerv2-large-p14-res336_clip-kinetics710-pre_u32_kinetics700-rgb |
82.10 |
96.0 |
|||
vit-base-p16_videomae-k400-pre_16x4x1_kinetics-400 |
81.30 |
95.0 |
|||
vit-large-p16_videomae-k400-pre_16x4x1_kinetics-400 |
85.30 |
96.7 |
|||
vit-small-p16_videomaev2-vit-g-dist-k710-pre_16x4x1_kinetics-400 |
83.60 |
96.3 |
|||
vit-base-p16_videomaev2-vit-g-dist-k710-pre_16x4x1_kinetics-400 |
86.60 |
97.3 |
|||
x3d_s_13x6x1_facebook-kinetics400-rgb |
3.79 |
2.97 |
73.30 |
||
x3d_m_16x5x1_facebook-kinetics400-rgb |
3.79 |
6.49 |
76.40 |
||
tsn_r18_8xb320-64x1x1-100e_kinetics400-audio-feature |
11.40 |
0.37 |
13.70 |
27.3 |
UCF101¶
模型 |
参数量 (M) |
Flops (G) |
Top-1 (%) |
Top-5 (%) |
Readme |
|---|---|---|---|---|---|
c3d_sports1m-pretrained_8xb30-16x1x1-45e_ucf101-rgb |
78.40 |
38.50 |
83.08 |
95.93 |
SthV2¶
模型 |
参数量 (M) |
Flops (G) |
Top-1 (%) |
Top 1 Accuracy (efficient) |
Top-5 (%) |
Top 5 Accuracy (efficient) |
Readme |
|---|---|---|---|---|---|---|---|
mvit-small-p244_k400-pre_16xb16-u16-100e_sthv2-rgb_infer |
68.10 |
91.00 |
|||||
mvit-small-p244_k400-pre_16xb16-u16-100e_sthv2-rgb |
34.40 |
64.00 |
68.20 |
91.30 |
|||
mvit-base-p244_u32_sthv2-rgb |
70.80 |
92.70 |
|||||
mvit-large-p244_u40_sthv2-rgb |
73.20 |
94.00 |
|||||
tin_imagenet-pretrained-r50_8xb6-1x1x8-40e_sthv2-rgb |
23.90 |
32.96 |
54.78 |
82.18 |
|||
trn_imagenet-pretrained-r50_8xb16-1x1x8-50e_sthv2-rgb |
42.94 |
51.20 |
47.65 |
78.42 |
76.27 |
||
tsm_imagenet-pretrained-r50_8xb16-1x1x8-50e_sthv2-rgb |
23.87 |
32.88 |
62.72 |
87.70 |
|||
tsm_imagenet-pretrained-r50_8xb16-1x1x16-50e_sthv2-rgb |
23.87 |
65.75 |
64.16 |
88.61 |
|||
tsm_imagenet-pretrained-r101_8xb16-1x1x8-50e_sthv2-rgb |
42.86 |
62.66 |
63.70 |
88.28 |
Kinetics-700¶
模型 |
参数量 (M) |
Flops (G) |
Top-1 (%) |
Top-5 (%) |
Readme |
|---|---|---|---|---|---|
slowonly_imagenet-pretrained-r50_16xb16-4x16x1-steplr-150e_kinetics700-rgb |
32.45 |
27.38 |
65.18 |
86.05 |
|
slowonly_imagenet-pretrained-r50_16xb16-8x8x1-steplr-150e_kinetics700-rgb |
32.45 |
54.75 |
66.93 |
87.47 |
|
swin-large-p244-w877_in22k-pre_16xb8-amp-32x2x1-30e_kinetics700-rgb |
197.00 |
604.00 |
75.92 |
92.72 |
|
uniformerv2-base-p16-res224_clip-pre_8xb32-u8_kinetics700-rgb |
75.90 |
92.90 |
|||
uniformerv2-base-p16-res224_clip-kinetics710-pre_8xb32-u8_kinetics700-rgb |
76.30 |
92.90 |
|||
uniformerv2-large-p14-res224_clip-kinetics710-pre_u8_kinetics700-rgb |
80.80 |
95.20 |
|||
uniformerv2-large-p14-res224_clip-kinetics710-pre_u16_kinetics700-rgb |
81.20 |
95.60 |
|||
uniformerv2-large-p14-res224_clip-kinetics710-pre_u32_kinetics700-rgb |
81.40 |
95.70 |
Kinetics-710¶
模型 |
参数量 (M) |
Flops (G) |
Top-1 (%) |
Top-5 (%) |
Readme |
|---|---|---|---|---|---|
slowonly_imagenet-pretrained-r50_32xb8-8x8x1-steplr-150e_kinetics710-rgb |
32.45 |
54.75 |
72.39 |
90.60 |
|
swin-small-p244-w877_in1k-pre_32xb4-amp-32x2x1-30e_kinetics710-rgb |
197.00 |
604.00 |
76.90 |
92.96 |
SthV1¶
模型 |
参数量 (M) |
Flops (G) |
Top-1 (%) |
Top 1 Accuracy (efficient) |
Top-5 (%) |
Top 5 Accuracy (efficient) |
Readme |
|---|---|---|---|---|---|---|---|
tanet_imagenet-pretrained-r50_8xb8-1x1x8-50e_sthv1-rgb |
25.10 |
43.10 |
49.71 |
46.98 |
77.43 |
75.75 |
|
tanet_imagenet-pretrained-r50_8xb6-1x1x16-50e_sthv1-rgb |
25.10 |
86.10 |
50.95 |
48.24 |
79.28 |
78.16 |
|
tin_imagenet-pretrained-r50_8xb6-1x1x8-40e_sthv1-rgb |
23.90 |
32.96 |
38.68 |
68.55 |
|||
tpn-tsm_imagenet-pretrained-r50_8xb8-1x1x8-150e_sthv1-rgb |
82.45 |
54.20 |
51.87 |
79.67 |
|||
trn_imagenet-pretrained-r50_8xb16-1x1x8-50e_sthv1-rgb |
42.94 |
33.65 |
31.6 |
62.22 |
60.15 |
Kinetics-600¶
模型 |
参数量 (M) |
Flops (G) |
Top-1 (%) |
Top-5 (%) |
Readme |
|---|---|---|---|---|---|
uniformerv2-base-p16-res224_clip-kinetics710-pre_8xb32-u8_kinetics600-rgb |
86.40 |
97.30 |
|||
uniformerv2-large-p14-res224_clip-kinetics710-pre_u8_kinetics600-rgb |
89.00 |
98.30 |
|||
uniformerv2-large-p14-res224_clip-kinetics710-pre_u16_kinetics600-rgb |
89.40 |
98.30 |
|||
uniformerv2-large-p14-res224_clip-kinetics710-pre_u32_kinetics600-rgb |
89.20 |
98.30 |
|||
uniformerv2-large-p14-res336_clip-kinetics710-pre_u32_kinetics600-rgb |
89.80 |
98.50 |
Moments in Time V1¶
模型 |
参数量 (M) |
Flops (G) |
Top-1 (%) |
Top-5 (%) |
Readme |
|---|---|---|---|---|---|
uniformerv2-base-p16-res224_clip-kinetics710-kinetics-k400-pre_16xb32-u8_mitv1-rgb |
42.30 |
71.50 |
|||
uniformerv2-large-p16-res224_clip-kinetics710-kinetics-k400-pre_u8_mitv1-rgb |
47.00 |
76.10 |
|||
uniformerv2-large-p16-res336_clip-kinetics710-kinetics-k400-pre_u8_mitv1-rgb |
47.70 |
76.80 |
时空行为检测¶
AVA v2.1¶
模型 |
参数量 (M) |
Flops (G) |
mAP |
Readme |
|---|---|---|---|---|
slowfast-acrn_kinetics400-pretrained-r50_8xb8-8x8x1-cosine-10e_ava21-rgb |
27.65 |
|||
slowonly-lfb-nl_kinetics400-pretrained-r50_8xb12-4x16x1-20e_ava21-rgb |
24.11 |
|||
slowonly-lfb-max_kinetics400-pretrained-r50_8xb12-4x16x1-20e_ava21-rgb |
22.15 |
|||
slowfast_kinetics400-pretrained-r50_8xb16-4x16x1-20e_ava21-rgb |
24.32 |
|||
slowfast_kinetics400-pretrained-r50-context_8xb16-4x16x1-20e_ava21-rgb |
25.34 |
|||
slowfast_kinetics400-pretrained-r50_8xb8-8x8x1-20e_ava21-rgb |
25.80 |
|||
slowonly_kinetics400-pretrained-r50_8xb16-4x16x1-20e_ava21-rgb |
20.72 |
|||
slowonly_kinetics700-pretrained-r50_8xb16-4x16x1-20e_ava21-rgb |
22.77 |
|||
slowonly_kinetics400-pretrained-r50-nl_8xb16-4x16x1-20e_ava21-rgb |
21.55 |
|||
slowonly_kinetics400-pretrained-r50-nl_8xb16-8x8x1-20e_ava21-rgb |
23.77 |
|||
slowonly_kinetics400-pretrained-r101_8xb16-8x8x1-20e_ava21-rgb |
24.83 |
AVA v2.2¶
模型 |
参数量 (M) |
Flops (G) |
mAP |
Readme |
|---|---|---|---|---|
slowfast-acrn_kinetics400-pretrained-r50_8xb8-8x8x1-cosine-10e_ava22-rgb |
27.71 |
|||
slowfast_kinetics400-pretrained-r50_8xb6-8x8x1-cosine-10e_ava22-rgb |
25.90 |
|||
slowfast_kinetics400-pretrained-r50-temporal-max_8xb6-8x8x1-cosine-10e_ava22-rgb |
26.41 |
|||
slowfast_r50-k400-pre-temporal-max-focal-alpha3-gamma1_8xb6-8x8x1-cosine-10e_ava22-rgb |
26.65 |
|||
vit-base-p16_videomae-k400-pre_8xb8-16x4x1-20e-adamw_ava-kinetics-rgb |
33.60 |
|||
vit-large-p16_videomae-k400-pre_8xb8-16x4x1-20e-adamw_ava-kinetics-rgb |
38.70 |
MultiSports¶
模型 |
参数量 (M) |
Flops (G) |
f-mAP |
Readme |
|---|---|---|---|---|
slowfast_kinetics400-pretrained-r50_8xb16-4x16x1-8e_multisports-rgb |
36.88 |
|||
slowonly_kinetics400-pretrained-r50_8xb16-4x16x1-8e_multisports-rgb |
26.40 |
骨骼点行为识别¶
NTU60-XSub-2D¶
模型 |
参数量 (M) |
Flops (G) |
Top-1 (%) |
Readme |
|---|---|---|---|---|
2s-agcn_8xb16-joint-u100-80e_ntu60-xsub-keypoint-2d |
3.50 |
4.40 |
88.60 |
|
2s-agcn_8xb16-bone-u100-80e_ntu60-xsub-keypoint-2d |
3.50 |
4.40 |
91.59 |
|
2s-agcn_8xb16-joint-motion-u100-80e_ntu60-xsub-keypoint-2d |
3.50 |
4.40 |
88.02 |
|
2s-agcn_8xb16-bone-motion-u100-80e_ntu60-xsub-keypoint-2d |
3.50 |
4.40 |
88.82 |
|
stgcn_8xb16-joint-u100-80e_ntu60-xsub-keypoint-2d |
3.10 |
3.80 |
88.95 |
|
stgcn_8xb16-bone-u100-80e_ntu60-xsub-keypoint-2d |
3.10 |
3.80 |
91.69 |
|
stgcn_8xb16-joint-motion-u100-80e_ntu60-xsub-keypoint-2d |
3.10 |
3.80 |
86.90 |
|
stgcn_8xb16-bone-motion-u100-80e_ntu60-xsub-keypoint-2d |
3.10 |
3.80 |
87.86 |
|
stgcnpp_8xb16-joint-u100-80e_ntu60-xsub-keypoint-2d |
1.39 |
1.95 |
89.29 |
|
stgcnpp_8xb16-bone-u100-80e_ntu60-xsub-keypoint-2d |
1.39 |
1.95 |
92.30 |
|
stgcnpp_8xb16-joint-motion-u100-80e_ntu60-xsub-keypoint-2d |
1.39 |
1.95 |
87.30 |
|
stgcnpp_8xb16-bone-motion-u100-80e_ntu60-xsub-keypoint-2d |
1.39 |
1.95 |
88.76 |
NTU60-XSub-3D¶
模型 |
参数量 (M) |
Flops (G) |
Top-1 (%) |
Readme |
|---|---|---|---|---|
2s-agcn_8xb16-joint-u100-80e_ntu60-xsub-keypoint-3d |
3.50 |
6.50 |
88.26 |
|
2s-agcn_8xb16-bone-u100-80e_ntu60-xsub-keypoint-3d |
3.50 |
6.50 |
89.22 |
|
2s-agcn_8xb16-joint-motion-u100-80e_ntu60-xsub-keypoint-3d |
3.50 |
6.50 |
86.73 |
|
2s-agcn_8xb16-bone-motion-u100-80e_ntu60-xsub-keypoint-3d |
3.50 |
6.50 |
86.41 |
|
stgcn_8xb16-joint-u100-80e_ntu60-xsub-keypoint-3d |
3.10 |
5.70 |
88.11 |
|
stgcn_8xb16-bone-u100-80e_ntu60-xsub-keypoint-3d |
3.10 |
5.70 |
88.76 |
|
stgcn_8xb16-joint-motion-u100-80e_ntu60-xsub-keypoint-3d |
3.10 |
5.70 |
86.06 |
|
stgcn_8xb16-bone-motion-u100-80e_ntu60-xsub-keypoint-3d |
3.10 |
5.70 |
85.49 |
|
stgcnpp_8xb16-joint-u100-80e_ntu60-xsub-keypoint-3d |
1.40 |
2.96 |
89.14 |
|
stgcnpp_8xb16-bone-u100-80e_ntu60-xsub-keypoint-3d |
1.40 |
2.96 |
90.21 |
|
stgcnpp_8xb16-joint-motion-u100-80e_ntu60-xsub-keypoint-3d |
1.40 |
2.96 |
86.67 |
|
stgcnpp_8xb16-bone-motion-u100-80e_ntu60-xsub-keypoint-3d |
1.40 |
2.96 |
87.45 |
FineGYM¶
模型 |
参数量 (M) |
Flops (G) |
mean Top 1 Accuracy |
Readme |
|---|---|---|---|---|
slowonly_r50_8xb16-u48-240e_gym-keypoint |
2.00 |
20.60 |
93.50 |
|
slowonly_r50_8xb16-u48-240e_gym-limb |
2.00 |
20.60 |
93.60 |
NTU60-XSub¶
模型 |
参数量 (M) |
Flops (G) |
Top-1 (%) |
Readme |
|---|---|---|---|---|
slowonly_r50_8xb16-u48-240e_ntu60-xsub-keypoint |
2.00 |
20.60 |
93.60 |
|
slowonly_r50_8xb16-u48-240e_ntu60-xsub-limb |
2.00 |
20.60 |
93.50 |
HMDB51¶
模型 |
参数量 (M) |
Flops (G) |
Top-1 (%) |
Readme |
|---|---|---|---|---|
slowonly_kinetics400-pretrained-r50_8xb16-u48-120e_hmdb51-split1-keypoint |
3.00 |
14.60 |
69.60 |
UCF101¶
模型 |
参数量 (M) |
Flops (G) |
Top-1 (%) |
Readme |
|---|---|---|---|---|
slowonly_kinetics400-pretrained-r50_8xb16-u48-120e_ucf101-split1-keypoint |
3.10 |
14.60 |
86.80 |
Kinetic400¶
模型 |
参数量 (M) |
Flops (G) |
Top-1 (%) |
Readme |
|---|---|---|---|---|
slowonly_r50_8xb32-u48-240e_k400-keypoint |
3.20 |
19.10 |
47.40 |
NTU120-XSub-2D¶
模型 |
参数量 (M) |
Flops (G) |
Top-1 (%) |
Readme |
|---|---|---|---|---|
stgcn_8xb16-joint-u100-80e_ntu120-xsub-keypoint-2d |
3.10 |
3.80 |
83.19 |
|
stgcn_8xb16-bone-u100-80e_ntu120-xsub-keypoint-2d |
3.10 |
3.80 |
83.36 |
|
stgcn_8xb16-joint-motion-u100-80e_ntu120-xsub-keypoint-2d |
3.10 |
3.80 |
78.87 |
|
stgcn_8xb16-bone-motion-u100-80e_ntu120-xsub-keypoint-2d |
3.10 |
3.80 |
79.55 |
NTU120-XSub-3D¶
模型 |
参数量 (M) |
Flops (G) |
Top-1 (%) |
Readme |
|---|---|---|---|---|
stgcn_8xb16-joint-u100-80e_ntu120-xsub-keypoint-3d |
3.10 |
5.70 |
82.15 |
|
stgcn_8xb16-bone-u100-80e_ntu120-xsub-keypoint-3d |
3.10 |
5.70 |
84.28 |
|
stgcn_8xb16-joint-motion-u100-80e_ntu120-xsub-keypoint-3d |
3.10 |
5.70 |
78.93 |
|
stgcn_8xb16-bone-motion-u100-80e_ntu120-xsub-keypoint-3d |
3.10 |
5.70 |
80.02 |
视频检索¶
MSRVTT¶
模型 |
参数量 (M) |
Flops (G) |
MdR |
MnR |
Recall@1 |
Recall@10 |
Recall@5 |
Readme |
|---|---|---|---|---|---|---|---|---|
clip4clip_vit-base-p32-res224-clip-pre_8xb16-u12-5e_msrvtt-9k-rgb |
2.00 |
16.80 |
43.10 |
78.90 |
69.40 |
时序动作定位¶
ActivityNet v1.3¶
模型 |
参数量 (M) |
Flops (G) |
AR@1 |
AR@10 |
AR@100 |
AR@5 |
AUC |
Readme |
|---|---|---|---|---|---|---|---|---|
bmn_2xb8-400x100-9e_activitynet-feature |
32.89 |
56.64 |
75.29 |
49.43 |
67.25 |
|||
bsn_400x100_1xb16_20e_activitynet_feature (cuhk_mean_100) |
32.71 |
55.28 |
74.27 |
48.43 |
66.26 |