Temporal Action Localization (Verb) on EPIC-Kitchens 100 (test)
34mAP (tIoU=0.1)ActionVLM-L-3B
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| ActionVLM-L-3BVision=VideoMAEv1-L, E2E=true2026.01 | 34 | 33.3 | 31.4 | 28.8 | 24.8 | 30.5 | |
| AdaTADVision=VideoMAEv1-L, E2E=true2026.01 | 33.1 | 32.2 | 30.4 | 27.5 | 23.1 | 29.3 | |
| ActionFormerVision=VideoMAEv1-L, E2E=false2026.01 | 32.7 | 31.6 | 29.1 | 26.7 | 23.6 | 28.7 | |
| TriDetBackbone=SlowFast2023.09 | 28.6 | 27.4 | 26.1 | 24.2 | 20.8 | 25.4 | |
| TemporalMaxer2023.03 | 27.8 | 26.6 | 25.3 | 23.1 | 19.9 | 24.5 | |
| ActionFormerVision=SlowFast, E2E=false2026.01 | 27.7 | 26.8 | 25.6 | 24.4 | 20.5 | 24.9 | |
| ActionFormer2023.03 | 26.6 | 25.4 | 24.2 | 22.3 | 19.1 | 23.5 | |
| ActionFormerBackbone=SlowFast2023.09 | 26.6 | 25.4 | 24.2 | 22.3 | 19.1 | 23.5 | |
| G-TAD2023.03 | 12.1 | 11 | 9.4 | 8.1 | 6.5 | 9.4 | |
| G-TADBackbone=SlowFast2023.09 | 12.1 | 11 | 9.4 | 8.1 | 6.5 | 9.4 | |
| BMN2023.03 | 10.8 | 9.8 | 8.4 | 7.1 | 5.6 | 8.4 | |
| BMNBackbone=SlowFast2023.09 | 10.8 | 8.8 | 8.4 | 7.1 | 5.6 | 8.4 |