Classification on VB100 (test)
18.2Accuracy (%)CLIP
Evaluation Results
| Method | Links | |
|---|---|---|
| CLIPModality=Vision, Protocol=Zero-shot, Frames=19 uniformly sampled2025.11 | 18.2 | |
| Tri-CLIPModality=Vision, Protocol=Zero-shot, Frames=19 uniformly sampled2025.11 | 17.7 | |
| ConFuModality=Vision, Protocol=Zero-shot, Frames=19 uniformly sampled2025.11 | 16.7 | |
| ConFuModality=Audio+Visual fusion, Protocol=Zero-shot2025.11 | 16.4 | |
| SymileModality=Audio+Visual fusion, Protocol=Zero-shot2025.11 | 13.2 | |
| GRAMModality=Vision, Protocol=Zero-shot, Frames=19 uniformly sampled2025.11 | 12.9 | |
| TRIANGLEModality=Audio+Visual fusion, Protocol=Zero-shot2025.11 | 8.7 | |
| GRAMModality=Audio+Visual fusion, Protocol=Zero-shot2025.11 | 7.6 | |
| CLIPModality=Audio, Protocol=Zero-shot2025.11 | 4.2 | |
| Tri-CLIPModality=Audio, Protocol=Zero-shot2025.11 | 3.9 | |
| ConFuModality=Audio, Protocol=Zero-shot2025.11 | 3.4 | |
| GRAMModality=Audio, Protocol=Zero-shot2025.11 | 1.3 |