Scientific Image Quality Assessment Understanding on SIQA-U
60.5Scientific CompletenessSIQA-Judger
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| SIQA-JudgerTraining=trained2026.03 | 60.5 | 54.6 | 52.5 | 57.1 | 57.4 | 83 | 29 | 56.3 | |
| InternVL3.5-241BAccess=Open-Source2026.03 | 47.9 | 46.9 | 46.1 | 48.4 | 46.5 | 68.6 | 27.1 | 47.3 | |
| LLaMA-3.2-90B-VisionAccess=Open-Source2026.03 | 46.8 | 44.4 | 47.2 | 44.4 | 48.3 | 62 | 28.6 | 45.9 | |
| Qwen3-VL-235BAccess=Open-Source2026.03 | 45.9 | 45.1 | 48.5 | 45.1 | 45.9 | 66 | 27 | 45.8 | |
| GLM-4.6vAccess=Open-Source2026.03 | 45.8 | 50.6 | 44.3 | 45.9 | 44.8 | 67.6 | 28.5 | 46.8 | |
| Gemini-2.5-ProAccess=Closed-Source2026.03 | 44.8 | 48.2 | 47.9 | 48.2 | 46.2 | 70 | 25.6 | 46.7 | |
| GPT-5Access=Closed-Source2026.03 | 44.1 | 45.5 | 47.9 | 45.5 | 44.8 | 67.7 | 24.3 | 45 | |
| Doubao-1.5-Vision-ProAccess=Closed-Source2026.03 | 43.7 | 42.9 | 48.6 | 42.9 | 47.8 | 64.2 | 24.1 | 44.9 | |
| Claude-sonnet-4.5Access=Closed-Source2026.03 | 43 | 50.7 | 50.8 | 50.7 | 43.5 | 71.6 | 30 | 47.8 | |
| GPT-4oAccess=Closed-Source2026.03 | 42.7 | 42.1 | 47.7 | 42.1 | 42.6 | 61.2 | 28.3 | 43.6 | |
| O3Access=Closed-Source2026.03 | 42.4 | 43.4 | 46.5 | 43.4 | 41.7 | 64.8 | 25.5 | 43.5 | |
| GPT-3.5-turboAccess=Closed-Source2026.03 | 39.9 | 46 | 45.9 | 46 | 44.4 | 59.5 | 27.1 | 43.2 | |
| Random Guess2026.03 | 36.5 | 33.6 | 33.9 | 36.1 | 50 | 25 | 25 | 35 | |
| DeepSeek-VL2Access=Open-Source2026.03 | 33 | 42.1 | 43.3 | 42.1 | 39.6 | 50.3 | 27.3 | 38.8 |