Visual Quality Grounding on IQA-Spider
2.41GPT-4V Grounding ScoreIQA-Spider
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| IQA-SpiderBackbone=Qwen3-VL2026.05 | 2.41 | — | — | — | — | |
| IQA-SpiderBackbone=Phi-3.5-Vision2026.05 | 2.29 | — | — | — | — | |
| IQA-SpiderBackbone=Qwen2.5-VL2026.05 | 2.13 | — | — | — | — | |
| Qwen2.5-VLBackbone=7B2026.05 | 1.25 | — | — | — | — | |
| Q-Instruct2026.05 | 0.93 | — | — | — | — | |
| Qwen3-VLBackbone=7B2026.05 | 0.7 | — | — | — | — | |
| Phi-3.5-Vision2026.05 | 0.68 | — | — | — | — | |
| Co-Instruct2026.05 | 0.52 | — | — | — | — | |
| DepictQA-Wild2026.05 | 0.24 | — | — | — | — | |
| IQA-SpiderBase Model=Phi-3.5-Vision2026.05 | — | 0.375 | 0.354 | 0.342 | 0.364 | |
| IQA-SpiderBase Model=Qwen2.5-VL2026.05 | — | 0.402 | 0.363 | 0.341 | 0.381 | |
| IQA-SpiderBase Model=Qwen3-VL2026.05 | — | 0.439 | 0.396 | 0.313 | 0.408 | |
| LISAModel size=7B2026.05 | — | 0.063 | 0.101 | 0.086 | 0.078 | |
| Q-Groundreproduction=true2026.05 | — | 0.192 | 0.176 | 0.202 | 0.19 | |
| Q-Instructparadigm=text-to-point2026.05 | — | 0.243 | 0.315 | 0.229 | 0.264 | |
| Sa2VAModel size=8B2026.05 | — | 0.186 | 0.194 | 0.209 | 0.192 |