Action Recognition on Charades v1 (val)
47.7mAPMViT-B-24
Evaluation Results
| Method | Links | |
|---|---|---|
| MViT-B-24pretrain=K600, Temporal Resolution=32x3, FLOPs x views=236x3x10, Param=53.02021.04 | 47.7 | |
| X3D-XLpretrain=K600, FLOPs x views=48.4x3x10, Param=11.02021.04 | 47.1 | |
| MViT-Bpretrain=K600, Temporal Resolution=32x3, FLOPs x views=170x3x10, Param=36.42021.04 | 47.1 | |
| MViT-Bpretrain=K400, Temporal Resolution=64x3, FLOPs x views=455x3x10, Param=36.42021.04 | 46.3 | |
| SlowFast R101+NLpretrain=K600, Temporal Resolution=16x8, FLOPs x views=234x3x10, Param=59.92021.04 | 45.2 | |
| MViT-Bpretrain=K400, Temporal Resolution=32x3, FLOPs x views=170x3x10, Param=36.42021.04 | 44.3 | |
| MViT-Bpretrain=K600, Temporal Resolution=16x4, FLOPs x views=70.5x3x10, Param=36.42021.04 | 43.9 | |
| X3D-XLpretrain=K400, FLOPs x views=48.4x3x10, Param=11.02021.04 | 43.4 | |
| LFB +NLpretrain=IN-1K+K400, FLOPs x views=529x3x10, Param=1222021.04 | 42.5 | |
| SlowFast 101+NLpretrain=K400, Temporal Resolution=16x8, FLOPs x views=234x3x10, Param=59.92021.04 | 42.5 | |
| Timeceptionpretrain=IN-1K+K4002021.04 | 41.1 | |
| MViT-Bpretrain=K400, Temporal Resolution=16x4, FLOPs x views=70.5x3x10, Param=36.42021.04 | 40 | |
| STRG +NLpretrain=IN-1K+K400, FLOPs x views=630x3x10, Param=58.32021.04 | 39.7 | |
| SlowFast 50pretrain=K400, Temporal Resolution=8x8, FLOPs x views=65.7x3x10, Param=34.02021.04 | 38 | |
| Nonlocalpretrain=IN-1K+K400, FLOPs x views=544x3x10, Param=54.32021.04 | 37.5 |