Semantic Role Labeling on VidSitu (test)
83.68CIDErHuman
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Human2026.04 | 83.68 | 71.77 | 70.6 | |
| CineMECBackbone=YS + SF2026.04 | 75.73 | 55.06 | 51.16 | |
| VideoWhispererBackbone=YS + SF2026.04 | 73.21 | 48.61 | 48.36 | |
| VideoWhispererBackbone=FR + SF2026.04 | 68.04 | 48.77 | 44.57 | |
| VidSitu-SlowFastBackbone=SF2026.04 | 47.25 | 50.88 | 33.5 |