Text-to-image synthesis evaluation (Human Correlation - Overall) on PaintSkills
0.6437Kendall's TauLLMScore
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| LLMScoreHuman ranking criteria=Error Counting2023.05 | 0.6437 | 0.7325 | |
| BLIP-ITCHuman ranking criteria=Error Counting2023.05 | 0.6178 | 0.6511 | |
| LLMScore2023.05 | 0.56 | 0.6853 | |
| LLMScoreHuman ranking criteria=Overall2023.05 | 0.56 | 0.6853 | |
| NegCLIP2023.05 | 0.4911 | 0.6313 | |
| NegCLIPHuman ranking criteria=Overall2023.05 | 0.4911 | 0.6313 | |
| BLIP-ITM2023.05 | 0.4755 | 0.6214 | |
| BLIP-ITMHuman ranking criteria=Overall2023.05 | 0.4755 | 0.6214 | |
| BLIP-ITC2023.05 | 0.4743 | 0.5864 | |
| BLIP-ITCHuman ranking criteria=Overall2023.05 | 0.4743 | 0.5864 | |
| CLIP2023.05 | 0.4715 | 0.5869 | |
| CLIPHuman ranking criteria=Overall2023.05 | 0.4715 | 0.5869 | |
| BLIP-ITMHuman ranking criteria=Error Counting2023.05 | 0.4622 | 0.4997 | |
| NegCLIPHuman ranking criteria=Error Counting2023.05 | 0.4563 | 0.4908 | |
| DescCLIPtype=LLMScore variant2023.05 | 0.3825 | 0.4917 | |
| CLIPHuman ranking criteria=Error Counting2023.05 | 0.3052 | 0.2891 | |
| CapCLIPtype=LLMScore variant2023.05 | 0.3035 | 0.3751 | |
| CapMETEORtype=LLMScore variant2023.05 | 0.2315 | 0.3056 | |
| DescMETEORtype=LLMScore variant2023.05 | 0.0678 | 0.0827 |