Correlation with human judgments on Flickr8k Expert (tau_b)
63.4Tau-b CorrelationVC-Inspector-7B
Evaluation Results
| Method | Links | |
|---|---|---|
| VC-Inspector-7BReference Setting=Reference-free2025.09 | 63.4 | |
| VC-Inspector-3BReference Setting=Reference-free2025.09 | 59.9 | |
| HICE-SReference Setting=Reference-based2025.09 | 57.2 | |
| PolosReference Setting=Reference-based2025.09 | 56.4 | |
| HICE-SReference Setting=Reference-free2025.09 | 55.9 | |
| PAC-SReference Setting=Reference-based2025.09 | 55.5 | |
| PAC-S++Reference Setting=Reference-based, Backbone=ViT-B2025.09 | 55.3 | |
| PAC-S++Reference Setting=Reference-free, Backbone=ViT-B2025.09 | 54.1 | |
| PAC-SReference Setting=Reference-free2025.09 | 53.9 | |
| FLEURReference Setting=Reference-free2025.09 | 52.7 | |
| CLIPScoreReference Setting=Reference-based2025.09 | 52.6 | |
| SPICEReference Setting=Reference-based2025.09 | 51.7 | |
| FLEURReference Setting=Reference-based2025.09 | 51.6 | |
| CLIPScoreReference Setting=Reference-free2025.09 | 51.1 | |
| CIDErReference Setting=Reference-based2025.09 | 43.6 | |
| METEORReference Setting=Reference-based2025.09 | 41.5 | |
| BLEU_1Reference Setting=Reference-based2025.09 | 32.2 | |
| ROUGEReference Setting=Reference-based2025.09 | 32.1 | |
| BLEU_4Reference Setting=Reference-based2025.09 | 30.6 |