Scientific Image Quality Assessment on SIQA-S 1.0 (test)
0.857Perception SRCCSIQA-Judger
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| SIQA-JudgerEvaluation Protocol=trained2026.03 | 0.857 | 0.881 | 0.915 | 0.937 | 0.886 | 0.909 | |
| HyperIQAModel Category=Natural IQA, Evaluation Protocol=trained2026.03 | 0.773 | 0.783 | 0.897 | 0.895 | 0.835 | 0.839 | |
| Q-AlignModel Category=Natural IQA, Evaluation Protocol=zero-shot2026.03 | 0.749 | 0.762 | 0.285 | 0.4 | 0.517 | 0.581 | |
| CLIP-IQA+Model Category=Natural IQA, Evaluation Protocol=trained2026.03 | 0.724 | 0.676 | 0.862 | 0.801 | 0.793 | 0.741 | |
| GPT-4oModel Category=Closed-Source MLLM, Evaluation Protocol=zero-shot2026.03 | 0.694 | 0.699 | 0.836 | 0.85 | 0.765 | 0.775 | |
| Qwen3-VL-235BModel Category=Open-Source MLLM, Evaluation Protocol=zero-shot2026.03 | 0.629 | 0.678 | 0.862 | 0.831 | 0.746 | 0.755 | |
| InternVL3.5-241BModel Category=Open-Source MLLM, Evaluation Protocol=zero-shot2026.03 | 0.6 | 0.638 | 0.835 | 0.835 | 0.717 | 0.736 | |
| Doubao-1.5-Vision-ProModel Category=Closed-Source MLLM, Evaluation Protocol=zero-shot2026.03 | 0.596 | 0.669 | 0.799 | 0.845 | 0.698 | 0.757 | |
| Llama-3.2-90B-VisionModel Category=Open-Source MLLM, Evaluation Protocol=zero-shot2026.03 | 0.568 | 0.611 | 0.482 | 0.58 | 0.525 | 0.596 | |
| CLIP-IQAModel Category=Natural IQA, Evaluation Protocol=zero-shot2026.03 | 0.496 | 0.52 | 0.362 | 0.435 | 0.429 | 0.478 | |
| GPT-3.5-turboModel Category=Closed-Source MLLM, Evaluation Protocol=zero-shot2026.03 | 0.48 | 0.425 | 0.699 | 0.665 | 0.589 | 0.545 | |
| NIQEModel Category=Natural IQA, Evaluation Protocol=zero-shot2026.03 | 0.345 | 0.235 | 0.447 | 0.41 | 0.396 | 0.322 |