Video Segmentation on YTSeg <10 min duration (test)
50.01F1 ScoreAudioSeg
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| AudioSegBackbone=Whisper Large2026.02 | 50.01 | 40.72 | 28.08 | |
| MiniSegInput=ASR_T + Features2026.02 | 44.77 | 38.22 | 29.77 | |
| MiniSegInput=Ref+ASR_T2026.02 | 44.06 | 36.98 | 30 | |
| Qwen3-OmniStrategy=ICL2026.02 | 41.8 | 36.12 | 35.49 | |
| Qwen2.5-OmniStrategy=LoRA2026.02 | 30.33 | 24.13 | 35.47 |