Referring Audio-Visual Segmentation on Ref-AVS 1.0 (Mix (S+U))
55Jaccard (J)SAM2-LOVE
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| SAM2-LOVEInput Modality=Audio + Text2025.06 | 55 | 62.1 | 58.5 | |
| EEMCInput Modality=Audio + Text2025.06 | 41.9 | 58.1 | 50 | |
| AVSegFormerInput Modality=Audio + Text2025.06 | 34.8 | 48.6 | 41.7 | |
| AVSegFormerInput Modality=Audio2025.06 | 30.9 | 43.7 | 37.3 | |
| ReferFormerInput Modality=Text + Audio2025.06 | 30.9 | 49.5 | 40.2 | |
| GAVSInput Modality=Audio + Text2025.06 | 29.4 | 49.8 | 39.6 | |
| AVSBenchInput Modality=Audio + Text2025.06 | 27.8 | 52.9 | 40.3 | |
| R2VOSInput Modality=Text + Audio2025.06 | 26.5 | 45.4 | 35.9 | |
| GAVSInput Modality=Audio2025.06 | 26.3 | 45.7 | 36 | |
| ReferFormerInput Modality=Text2025.06 | 26.2 | 49.1 | 37.6 | |
| AVSBenchInput Modality=Audio2025.06 | 24.5 | 48.3 | 36.4 | |
| R2VOSInput Modality=Text2025.06 | 22.6 | 37 | 29.8 |