Temporal Action Localization (Noun) on EPIC-Kitchens 100 (test)
32.8mAP (tIoU=0.1)ActionVLM-L-3B
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| ActionVLM-L-3BVision=VideoMAEv1-L, E2E=true2026.01 | 32.8 | 32.1 | 30.7 | 28.2 | 26.1 | 30 | |
| AdaTADVision=VideoMAEv1-L, E2E=true2026.01 | 32.4 | 31.6 | 30.1 | 27.4 | 24.6 | 29.3 | |
| ActionFormerVision=VideoMAEv1-L, E2E=false2026.01 | 31.3 | 29.7 | 27.2 | 25.3 | 21.3 | 26.9 | |
| TriDetBackbone=SlowFast2023.09 | 27.4 | 26.3 | 24.6 | 22.2 | 18.3 | 23.8 | |
| TemporalMaxer2023.03 | 26.3 | 25.2 | 23.5 | 21.3 | 17.6 | 22.8 | |
| ActionFormerVision=SlowFast, E2E=false2026.01 | 25.8 | 24.7 | 22.8 | 20.8 | 17.5 | 22.3 | |
| ActionFormer2023.03 | 25.2 | 24.1 | 22.7 | 20.5 | 17 | 21.9 | |
| ActionFormerBackbone=SlowFast2023.09 | 25.2 | 24.1 | 22.7 | 20.5 | 17 | 21.9 | |
| G-TAD2023.03 | 11 | 10 | 8.6 | 7 | 5.4 | 8.4 | |
| G-TADBackbone=SlowFast2023.09 | 11 | 10 | 8.6 | 7 | 5.4 | 8.4 | |
| BMN2023.03 | 10.3 | 8.3 | 6.2 | 4.5 | 3.4 | 6.5 | |
| BMNBackbone=SlowFast2023.09 | 10.3 | 8.3 | 6.2 | 4.5 | 3.4 | 6.5 |