Fine-grained classification on Cars-196
63.1cACCCLIP (zero-shot)
Evaluation results
| Method | Links | ||
|---|---|---|---|
| CLIP (zero-shot)protocol=zero-shot, ground_truth_class_names=known2025.12 | 63.1 | 66.3 | |
| FiNDRvocabulary_type=Automatically discovered, label_discovery_set=3 images per class, average_runs=102025.12 | 56.1 | 64.4 | |
| RAR2024.03 | 53.2 | 63.6 | |
| E-FineRvocabulary_type=Automatically discovered, label_discovery_set=3 images per class, average_runs=102025.12 | 51.2 | 64 | |
| FineRvocabulary_type=Automatically discovered, label_discovery_set=3 images per class, average_runs=102025.12 | 49.2 | 63.5 | |
| FineR2024.03 | 49.2 | 63.5 | |
| BLIP-2vocabulary_type=Automatically discovered, label_discovery_set=3 images per class, average_runs=102025.12 | 43.1 | 57.9 | |
| BLIP-22024.03 | 43.1 | 57.9 | |
| KMeansvocabulary_type=Index-based, without_vocabulary=true2025.12 | 30.6 | — | |
| CaSEDvocabulary_type=Pre-defined2025.12 | 26.9 | 41.4 | |
| CaSED2024.03 | 26.9 | 41.4 | |
| WordNetvocabulary_type=Pre-defined2025.12 | 18.3 | 33.3 | |
| WordNet+CLIP2024.03 | 18.3 | 33.3 | |
| CLIP-Sinkhornvocabulary_type=Index-based, without_vocabulary=true2025.12 | 18.1 | — | |
| DINO-Sinkhornvocabulary_type=Index-based, without_vocabulary=true2025.12 | 7.4 | — |