Multimodal Retrieval on MUStARD
81Recall@10 (Q: M23, T: M1)GRAM
Evaluation Results
| Method | Links | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| GRAM2025.11 | 81 | 61.7 | 24.8 | 67.2 | 24.8 | 67.8 | 27.9 | 5.1 | 31.2 | 35.2 | 60 | |
| ConFu2025.11 | 79.6 | 73.8 | 33.6 | 74.4 | 28.1 | 74.6 | 33.9 | 28 | 33.5 | 45.3 | 62.6 | |
| TRIANGLE2025.11 | 68.7 | — | — | — | — | 59.9 | — | — | 21.9 | — | 50.2 | |
| Symile2025.11 | 61.5 | — | — | — | — | 57 | — | — | 21.3 | — | 46.6 | |
| Trimodal CLIP2025.11 | — | 70.7 | 29.1 | 70.5 | 26.5 | — | 30.4 | 24.6 | — | 42 | — |