Action Recognition on FineGym Gym288
0.686Mean per-class AccuracyTAG-Head
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| TAG-HeadInput=R2026.04 | 0.686 | — | 0.922 | |
| PEVLInput=R + T + P2026.04 | 0.64 | — | 0.918 | |
| PGVTInput=R + P2026.04 | 0.636 | — | 0.91 | |
| VTCEInput=R + T2026.04 | 0.626 | — | 0.901 | |
| TQNInput=R2026.04 | 0.619 | — | 0.896 | |
| SlowFastInput=R2026.04 | 0.512 | — | 0.868 | |
| RSANet-R50Backbone=ResNet-502021.11 | 0.509 | — | — | |
| SELFYNet TSM R50#frame=82021.02 | 0.495 | — | — | |
| TSM-R50#frame=8, implementation=our impl.2021.02 | 0.479 | — | — | |
| TSM Two-Stream#frame=N/A2021.02 | 0.465 | — | — | |
| TSMTwo-stream2021.11 | 0.465 | — | — | |
| TSMInput=R2026.04 | 0.465 | — | 0.735 | |
| TRNmsInput=R2026.04 | 0.433 | — | 0.731 | |
| TSNInput=R2026.04 | 0.376 | — | 0.683 | |
| TSM-R50#frame=3, implementation=our impl.2021.02 | 0.353 | — | — | |
| TSM#frame=32021.02 | 0.348 | — | — | |
| TSM2021.11 | 0.348 | — | — | |
| TRN#frame=32021.02 | 0.331 | — | — | |
| TRN2021.11 | 0.331 | — | — | |
| I3DInput=R2026.04 | 0.282 | — | 0.667 | |
| I3D#frame=82021.02 | 0.279 | — | — | |
| I3D2021.11 | 0.279 | — | — | |
| NL I3D#frame=82021.02 | 0.271 | — | — | |
| TSN#frame=32021.02 | 0.265 | — | — | |
| LVMAE2024.11 | — | 0.928 | — | |
| MOSS-BBackbone=ViT-B/162026.04 | — | — | 0.702 | |
| MOSS-LBackbone=ViT-L/142026.04 | — | — | 0.711 | |
| RSANet2026.04 | — | — | 0.509 | |
| Side4Video-B*Backbone=ViT-B, Note=reproduced by current paper setup2026.04 | — | — | 0.691 | |
| StructViT-B-4-1Backbone=ViT-B2026.04 | — | — | 0.542 | |
| TCT2024.11 | — | 0.926 | — | |
| TQN2024.11 | — | 0.896 | — | |
| TQN2026.04 | — | — | 0.619 | |
| TSM2024.11 | — | 0.831 | — | |
| TSM2026.04 | — | — | 0.348 | |
| TSMtwo-stream2026.04 | — | — | 0.465 | |
| VT-CE2024.11 | — | 0.901 | — | |
| VT-CE2026.04 | — | — | 0.626 |