Audio-Visual Segmentation on AVSBench binary V2
79.09MJ ScoreAuralSAM2
Evaluation results
| Method | Links | ||
|---|---|---|---|
| AuralSAM2Backbones (audio & visual)=VGGish Hiera-l2025.06 | 79.09 | 86.84 | |
| AuralSAM2Backbones (audio & visual)=VGGish Hiera-b+2025.06 | 76.78 | 85.38 | |
| SAMA-AVSBackbones (audio & visual)=VGGish Hiera-b+, reimplementation=true2025.06 | 74.28 | 84.35 | |
| GAVSBackbones (audio & visual)=VGGish Hiera-b+, reimplementation=true2025.06 | 73.58 | 84.04 | |
| SAM (w/ AuralFuser)Backbones (audio & visual)=VGGish ViT-h2025.06 | 70.24 | 81.63 | |
| GAVSBackbones (audio & visual)=VGGish ViT-b2025.06 | 67.7 | 78.8 | |
| AVS-BiGenBackbones (audio & visual)=VGGish PVT-v22025.06 | 64.3 | 75.9 | |
| AL-RefBackbones (audio & visual)=Beats Hiera-l, zero-shot capability=true2025.06 | 59.2 | 66.2 |