Instruction-based Image Editing on GIER 50 user study samples
0.225L MetricJarvisArt
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| JarvisArtCategory=Tool-Calling Methods2026.06 | 0.225 | -0.328 | |
| Gemini-2.5-ProCategory=Tool-Calling Methods2026.06 | 0.225 | -0.42 | |
| GPT-Image-1Category=Generative Methods2026.06 | 0.218 | -0.34 | |
| GPT-4.1Category=Tool-Calling Methods2026.06 | 0.209 | -0.306 | |
| Qwen2.5-VL-7BCategory=Tool-Calling Methods2026.06 | 0.2 | -0.221 | |
| Qwen-Image-EditCategory=Generative Methods2026.06 | 0.184 | -0.132 | |
| IEACategory=Tool-Calling Methods2026.06 | 0.128 | 0.138 | |
| ReferenceType=Human Expert2026.06 | 0 | 1 |