DrawBench
Benchmarks
| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| Text-to-image generation | DrawBench | Latency (s)5.5 | 48 | |
| Text-to-Image Generation | DrawBench | Pick Score23.26 | 40 | |
| Text-to-image generation | Short-DrawBench 1k prompts Stable Diffusion v1.5 base (test) | R1 Score0.43 | 35 | |
| Text-to-Image Generation | DrawBench | HPSv231.58 | 27 | |
| Text-to-Image Generation | DrawBench | Aes.6.89 | 25 | |
| Text-to-Image Generation | DrawBench | PickScore23.8 | 23 | |
| text-to-image generation | DrawBench v1.0 (test) | Latency (s)2.55 | 22 | |
| Text-to-Image Generation | Drawbench | HPSV2.131.67 | 19 | |
| Text-to-image synthesis evaluation (Human Correlation - Overall) | DrawBench | Kendall's Tau0.223 | 19 | |
| Text-to-Image Generation | DrawBench | VQAScore0.901 | 18 | |
| Text-to-Image Generation | DrawBench Visual Text Rendering | PickScore23.68 | 17 | |
| Text-to-Image Generation | DrawBench FLUX.1 (dev) | IR0.9655 | 13 | |
| Text-to-Image Generation | DrawBench | VS8.847 | 12 | |
| Text-to-Image Generation | DrawBench | PickScore24.05 | 12 | |
| Image Generation | DrawBench | Aesthetic Score5.45 | 10 | |
| Text-to-Image Generation | DrawBench evaluated with Stable Diffusion 3.5 medium 1.0 (test) | IR0.97 | 10 | |
| Text-to-Image Generation | DrawBench | IR (Similarity Score)97 | 10 | |
| Text-to-Image Generation | DrawBench v1 (test) | VQAScore0.885 | 10 | |
| Text-to-image generation | DrawBench 512 x 512 resolution | ImageReward1.0092 | 10 | |
| Compositional Image Generation | DrawBench | Aesthetics Score5.44 | 9 | |
| Text-to-image generation | Drawbench | CLIP Score27.75 | 9 | |
| Text-to-image generation | DrawBench | ImgRwd1.01 | 9 | |
| Text-to-image synthesis evaluation (Human Correlation - Error Counting) | DrawBench | Kendall's Tau0.2125 | 9 | |
| Layout-to-Image Generation | DrawBench | Spatial Score60 | 8 | |
| Object Counting | DrawBench | Precision93.84 | 8 |
Showing 25 of 48 rows