Text-to-Image Retrieval on FLIR (test)
11.7Recall@1T-CLIP
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| T-CLIPSetting=Dual2026.05 | 11.7 | 36.4 | 51.1 | 72.8 | 85.1 | 93.2 | |
| Global LoRA2026.05 | 10.6 | 32.1 | 45.9 | 67.3 | 81.2 | 90.7 | |
| T-CLIPSetting=Global2026.05 | 10.5 | 34.8 | 49.2 | 71.2 | 83.9 | 92.5 | |
| DeCLIPSetting=Global2026.05 | 4.5 | 15.8 | 23.2 | 36.7 | 51 | 64.1 | |
| F-G LoRASetting=Fine-grained2026.05 | 3.5 | 12.7 | 20.6 | 36.9 | 52.8 | 70.2 | |
| CLIP-AdapterSetting=Global2026.05 | 3.1 | 13 | 21.7 | 38.1 | 52.9 | 70.6 | |
| T-CLIPSetting=Fine-grained2026.05 | 2 | 6.9 | 13 | 23.9 | 36.1 | 52.3 | |
| CLIP-AdapterSetting=Fine-grained2026.05 | 1.9 | 7.5 | 13.3 | 24.4 | 36.7 | 55.5 | |
| DeCLIPSetting=Fine-grained2026.05 | 1.5 | 7.3 | 11.4 | 21.8 | 32.4 | 47.4 | |
| CLIPSetting=Global2026.05 | 1.4 | 6.2 | 10.6 | 22.9 | 31.9 | 46.7 | |
| CLIPSetting=Fine-grained2026.05 | 0.5 | 1.9 | 3.2 | 9.3 | 15.8 | 25.3 |