Video Retrieval on AuroraCap-VideoDetailCaption Long Caption 1.0 (test)
75.7R@1ViLL-E
Evaluation results
| Method | Links | ||
|---|---|---|---|
| ViLL-EProtocol=Zero-shot, Caption Type=Long, Direction=T → V2026.04 | 75.7 | 92.4 | |
| ViLL-EProtocol=Zero-shot, Caption Type=Long, Direction=V → T2026.04 | 75.1 | 93.3 | |
| LongCLIPProtocol=Zero-shot, Caption Type=Long, Direction=V → T2026.04 | 74.5 | 92.8 | |
| LongCLIPProtocol=Zero-shot, Caption Type=Long, Direction=T → V2026.04 | 73.5 | 91.7 | |
| SigLIPProtocol=Zero-shot, Caption Type=Long, Direction=V → T2026.04 | 60.4 | 81 | |
| VidLAProtocol=Zero-shot, Caption Type=Long, Direction=V → T2026.04 | 55.9 | 72.7 | |
| SigLIPProtocol=Zero-shot, Caption Type=Long, Direction=T → V2026.04 | 51.7 | 72.7 | |
| VidLAProtocol=Zero-shot, Caption Type=Long, Direction=T → V2026.04 | 45.3 | 65.2 |