Image Retrieval on MUGE Retrieval
Metric: Mean Recall (higher is better)
Metric: Mean Recall (higher is better)
| # | Model↕ | Mean Recall▼ | Extra Data | Paper | Date↕ | Code |
|---|---|---|---|---|---|---|
| 1 | CN-CLIP (ViT-H/14) | 83.6 | No | Chinese CLIP: Contrastive Vision-Language Pretra... | 2022-11-02 | Code |
| 2 | CN-CLIP (ViT-L/14@336px) | 81.3 | No | Chinese CLIP: Contrastive Vision-Language Pretra... | 2022-11-02 | Code |
| 3 | CN-CLIP (ViT-L/14) | 80.1 | No | Chinese CLIP: Contrastive Vision-Language Pretra... | 2022-11-02 | Code |
| 4 | R2D2 (ViT-L/14) | 77.5 | No | CCMB: A Large-scale Chinese Cross-modal Benchmark | 2022-05-08 | Code |
| 5 | CN-CLIP (ViT-B/16) | 77.4 | No | Chinese CLIP: Contrastive Vision-Language Pretra... | 2022-11-02 | Code |
| 6 | Wukong (ViT-L/14) | 72.1 | No | Wukong: A 100 Million Large-scale Chinese Cross-... | 2022-02-14 | Code |
| 7 | CN-CLIP (RN50) | 69.2 | No | Chinese CLIP: Contrastive Vision-Language Pretra... | 2022-11-02 | Code |
| 8 | R2D2 (ViT-B) | 68.7 | No | CCMB: A Large-scale Chinese Cross-modal Benchmark | 2022-05-08 | Code |
| 9 | Wukong (ViT-B/32) | 61.2 | No | Wukong: A 100 Million Large-scale Chinese Cross-... | 2022-02-14 | Code |