Depth-to-Text Retrieval (via Vision pivot) on SUN to NYU (test)
38.7mAPIB+BB
Evaluation results
| Method | Links | |
|---|---|---|
| IB+BBBackbone=ImageBind, Variant=BrokenBind2026.02 | 38.7 | |
| LB+BBBackbone=LanguageBind, Variant=BrokenBind2026.02 | 36.8 | |
| LB+BBBackbone=LanguageBind, Variant=BrokenBind2026.02 | 32.1 | |
| IB+BBBackbone=ImageBind, Variant=BrokenBind2026.02 | 31.2 | |
| LB+FTBackbone=LanguageBind, Variant=Fine-tuned2026.02 | 15.2 | |
| IB+FTBackbone=ImageBind, Variant=Fine-tuned2026.02 | 15.1 | |
| LB+FTBackbone=LanguageBind, Variant=Fine-tuned2026.02 | 13.6 | |
| IBBackbone=ImageBind, Variant=Zero-shot2026.02 | 13.5 | |
| IB+FTBackbone=ImageBind, Variant=Fine-tuned2026.02 | 13.2 | |
| LBBackbone=LanguageBind, Variant=Zero-shot2026.02 | 12.6 | |
| LBBackbone=LanguageBind, Variant=Zero-shot2026.02 | 8.9 | |
| IBBackbone=ImageBind, Variant=Zero-shot2026.02 | 7.5 |