Video Segmentation on YTSeg 10–<30 min duration (test)
51.4F1 ScoreAudioSeg
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| AudioSegBackbone=Whisper Large2026.02 | 51.4 | 41.39 | 26.59 | |
| MiniSegInput=ASR_T + Features2026.02 | 44.4 | 37.13 | 28.22 | |
| Qwen3-OmniStrategy=ICL2026.02 | 40.63 | 34.41 | 30.77 | |
| MiniSegInput=Ref+ASR_T2026.02 | 40.56 | 33.17 | 29.71 | |
| Qwen2.5-OmniStrategy=LoRA2026.02 | 19.52 | 11.42 | 40.61 |