Action Recognition on GYM99
97Top-1 AccuracyPEVL
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| PEVLInput=R + T + P2026.04 | 97 | 91.8 | |
| PGVTInput=R + P2026.04 | 96.7 | 91.6 | |
| TAG-HeadInput=R2026.04 | 95.6 | 93.8 | |
| MOSS-LBackbone=ViT-L/142026.04 | 94.7 | — | |
| SlowFastInput=R2026.04 | 93.9 | 90.6 | |
| MOSS-BBackbone=ViT-B/162026.04 | 93.9 | — | |
| TQNInput=R2026.04 | 93.8 | 90.6 | |
| Side4Video-B*Backbone=ViT-B, Note=reproduced by current paper setup2026.04 | 92.3 | — | |
| VT-CE2026.04 | 91.4 | — | |
| TQN2026.04 | 90.6 | — | |
| VTCEInput=R + T2026.04 | 90.1 | 91.4 | |
| StructViT-B-4-1Backbone=ViT-B2026.04 | 89.5 | — | |
| TSMInput=R2026.04 | 88.4 | 81.2 | |
| TRNmsInput=R2026.04 | 87.8 | 80.2 | |
| RSANet2026.04 | 86.4 | — | |
| TSNInput=R2026.04 | 86 | 76.4 | |
| TSMtwo-stream2026.04 | 81.2 | — | |
| I3DInput=R2026.04 | 75.6 | 64.4 | |
| TSM2026.04 | 70.6 | — |