Video Reasoning Segmentation on VideoReasonSeg Overall (val)
55.4Seg J&FViLLa
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| ViLLaBackbone=InternVideo2-6B2024.07 | 55.4 | 49.9 | |
| ViLLaBackbone=InternVideo2-1B2024.07 | 51.2 | 43.1 | |
| VISABackbone=InternVideo2-6B2024.07 | 50.5 | 40.5 | |
| VideoLISA*Backbone=InternVideo2-6B2024.07 | 47.8 | 39.2 | |
| VISABackbone=Chat-UniVi-13B2024.07 | 47.6 | 39.6 | |
| VISABackbone=LLaVA-13B2024.07 | 47.4 | 37.9 | |
| VideoLISA*Backbone=InternVideo2-1B2024.07 | 45.3 | 36.7 | |
| VISABackbone=LLaVA-7B2024.07 | 43.1 | 35.2 | |
| LISABackbone=LLaVA-13B2024.07 | 39.2 | 32.9 | |
| LISABackbone=LLaVA-7B2024.07 | 38.4 | 32.1 | |
| LMPM2024.07 | 23.1 | — | |
| OnlineRefer2024.07 | 16.3 | — | |
| ReferFormer2024.07 | 16 | — | |
| MTTR2024.07 | 15.6 | — |