Infrared-Visible Image Fusion Quality Assessment on SemanticRT 5-expert pilot
0.561SRCCFuScore
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| FuScoreSeed Config=3-seed ensemble, Training Protocol=EVAFusion-trained2026.05 | 0.561 | 0.564 | 0.428 | 74.2 | |
| FuScoreSeed Config=single-seed, Training Protocol=EVAFusion-trained2026.05 | 0.547 | 0.551 | 0.413 | 73.1 | |
| ViT-L+MLPMethod Category=Deep Learning, Training Protocol=EVAFusion-trained2026.05 | 0.471 | 0.478 | 0.354 | 71.4 | |
| QabfMethod Category=Handcrafted2026.05 | 0.321 | 0.334 | 0.231 | 67.1 | |
| VIFMethod Category=Handcrafted2026.05 | 0.318 | 0.287 | 0.227 | 67.3 | |
| DeQA-ScoreMethod Category=Zero-shot VLM, Evaluation Weights=Zero-shot2026.05 | 0.314 | 0.301 | 0.234 | 68.2 | |
| Q-AlignMethod Category=Zero-shot VLM, Evaluation Weights=Zero-shot2026.05 | 0.296 | 0.284 | 0.219 | 69.8 | |
| MIMethod Category=Handcrafted2026.05 | 0.238 | 0.249 | 0.171 | 62.8 | |
| Qwen3-VL-8BMethod Category=Zero-shot VLM, Evaluation Weights=Zero-shot2026.05 | 0.142 | 0.157 | 0.103 | 60.2 |