Video Understanding on OmniVid
48.2OmniVid ScoreFull Token
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Full TokenBackbone=Qwen2.5-Omni 7B, Comp.↑ (%)=02026.05 | 48.2 | 100 | |
| ContextGuardBackbone=Qwen2.5-Omni 7B, Comp.↑ (%)=552026.05 | 48.2 | 100 | |
| FastVBackbone=Qwen2.5-Omni 7B, Comp.↑ (%)=502026.05 | 47.6 | 98.1 | |
| OmniZipBackbone=Qwen2.5-Omni 7B, Comp.↑ (%)=542026.05 | 47 | 97.3 | |
| ContextGuardBackbone=Video-SALMONN2+ 7B, Comp.↑ (%)=552026.05 | 47 | 100.5 | |
| Full TokenBackbone=Video-SALMONN2+ 3B, Comp.↑ (%)=02026.05 | 45.8 | 100 | |
| ContextGuardBackbone=Qwen2.5-Omni 3B, Comp.↑ (%)=612026.05 | 45.2 | 100.5 | |
| ContextGuardBackbone=Video-SALMONN2+ 3B, Comp.↑ (%)=562026.05 | 44.6 | 96.9 | |
| Full TokenBackbone=Qwen2.5-Omni 3B, Comp.↑ (%)=02026.05 | 44 | 100 | |
| FastVBackbone=Qwen2.5-Omni 3B, Comp.↑ (%)=502026.05 | 44 | 98.2 | |
| RandomBackbone=Qwen2.5-Omni 7B, Comp.↑ (%)=502026.05 | 43.4 | 95.4 | |
| Full TokenBackbone=Video-SALMONN2+ 7B, Comp.↑ (%)=02026.05 | 43.4 | 100 | |
| RandomBackbone=Qwen2.5-Omni 3B, Comp.↑ (%)=502026.05 | 42.8 | 95.7 | |
| OmniZipBackbone=Qwen2.5-Omni 3B, Comp.↑ (%)=542026.05 | 42.8 | 98.9 | |
| OmniZipBackbone=Video-SALMONN2+ 7B, Comp.↑ (%)=492026.05 | 42.8 | 95.7 | |
| FastVBackbone=Video-SALMONN2+ 7B, Comp.↑ (%)=502026.05 | 42.2 | 92.8 | |
| RandomBackbone=Video-SALMONN2+ 3B, Comp.↑ (%)=502026.05 | 42 | 90.6 | |
| OmniZipBackbone=Video-SALMONN2+ 3B, Comp.↑ (%)=492026.05 | 41.6 | 93.7 | |
| RandomBackbone=Video-SALMONN2+ 7B, Comp.↑ (%)=502026.05 | 39.5 | 90.7 | |
| FastVBackbone=Video-SALMONN2+ 3B, Comp.↑ (%)=502026.05 | 36.1 | 90.2 |