Action Recognition on Assembly101
47.51Action AccuracyTSM
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| TSMViews=All (8) Views2024.03 | 47.51 | 70.99 | 57.73 | — | |
| Video (Multimodal)Modality=Multimodal, Single-frame=true2026.01 | 43.6 | — | — | — | |
| TSMViews=Two Views2024.03 | 41.96 | 65.22 | 53.26 | — | |
| HandFormerViews=Single View + Pose2024.03 | 41.06 | 69.23 | 51.17 | — | |
| HandFormer-B/21×8Pose=true, RGB=true, GFLOPs=47.62024.03 | 41.06 | 69.23 | 51.17 | — | |
| MS-G3D + TSMPose=true, RGB=true, GFLOPs=66.22024.03 | 39.74 | 65.12 | 51.12 | — | |
| TSMViews=Single View + Egocentric2024.03 | 37.75 | 61.8 | 49.43 | — | |
| TSMViews=Single View2024.03 | 35.27 | 58.27 | 47.45 | — | |
| TSMPose=false, RGB=true, GFLOPs=33.02024.03 | 35.27 | 58.27 | 47.45 | — | |
| AAG_ARModality=RGB, Action History=true, Single-frame=true, Trainable Params (M)=24, Processed Frames=12026.01 | 34.19 | — | — | 58.3 | |
| RGBPoseConv3DPose=true, RGB=true, GFLOPs=58.02024.03 | 33.61 | 61.99 | 42.9 | — | |
| AAG_ARModality=RGB-D, Action History=true, Single-frame=true, Trainable Params (M)=24, Processed Frames=12026.01 | 32.77 | — | — | 56.19 | |
| HandFormer-B/21×8Pose=false, RGB=true, GFLOPs=33.02024.03 | 32.07 | 55.61 | 44.89 | — | |
| AAG+_ARModality=RGB-D, Action History=true, Single-frame=true, Trainable Params (M)=34, Processed Frames=12026.01 | 31.49 | — | — | 59.09 | |
| AAG+_ARModality=RGB, Action History=true, Single-frame=true, Trainable Params (M)=34, Processed Frames=12026.01 | 30.74 | — | — | 57.97 | |
| HandFormer-B/21×8Pose=true, RGB=false, GFLOPs=4.2+2024.03 | 28.8 | 65.33 | 36.28 | — | |
| MS-G3DPose=true, RGB=false, GFLOPs=21.22024.03 | 28.78 | 63.46 | 37.26 | — | |
| ISTA-NetPose=true, RGB=false, GFLOPs=35.22024.03 | 28.14 | 62.7 | 36.77 | — |