Verb Recognition on EPIC-Kitchens (S2: Unseen Kitchens)
31.92Shared AccX-MIC+DINO
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| X-MIC+DINOta (Temporal Attention)=true, Trained on=Ego4D (E4D)2024.03 | 31.92 | — | — | 6.38 | 10.63 | |
| X-MICta (Temporal Attention)=true, Trained on=Ego4D (E4D)2024.03 | 29.01 | — | — | 7 | 11.27 | |
| A5ta (Temporal Attention)=true, Trained on=Ego4D (E4D)2024.03 | 27.9 | — | — | 3.09 | 5.56 | |
| Vita-CLIPta (Temporal Attention)=true, Trained on=Ego4D (E4D)2024.03 | 27.22 | — | — | 4.11 | 7.14 | |
| CLIP-Adapter*ta (Temporal Attention)=true, Trained on=Ego4D (E4D)2024.03 | 25.29 | — | — | 1.23 | 2.35 | |
| CoOpta (Temporal Attention)=false, Trained on=Ego4D (E4D)2024.03 | 25.03 | — | — | 5.97 | 9.64 | |
| CoCoOpta (Temporal Attention)=false, Trained on=Ego4D (E4D)2024.03 | 24.34 | — | — | 0 | 0 | |
| CLIP-Adapterta (Temporal Attention)=false, Trained on=Ego4D (E4D)2024.03 | 21.48 | — | — | 3.09 | 5.4 | |
| ZS CLIPta (Temporal Attention)=false, Trained on=Ego4D (E4D)2024.03 | 12.32 | — | — | 4.32 | 6.4 | |
| Action BanksInput Modality=RGB, Number of frames=8, Number of crops=10, Resolution=full res2020.04 | — | 50.9 | 77.6 | — | — | |
| TRNInput Modality=RGB, Number of frames=8, Number of crops=102020.04 | — | 47.3 | 76.9 | — | — | |
| TRN-MultiscaleInput Modality=RGB, Number of frames=8, Number of crops=102020.04 | — | 46.9 | 75.2 | — | — | |
| TSMInput Modality=RGB, Number of frames=8, Number of crops=102020.04 | — | 43.5 | 73.9 | — | — | |
| TSM+W3Input Modality=RGB, Number of frames=8, Number of crops=102020.04 | — | 50.2 | 78 | — | — | |
| TSM+W3Input Modality=RGB, Number of frames=8, Number of crops=10, Resolution=full res2020.04 | — | 51.4 | 78.5 | — | — | |
| TSNInput Modality=RGB, Number of frames=8, Number of crops=102020.04 | — | 36.7 | 73.6 | — | — |