Image Retrieval On Inaturalist
Metrics
R@1
Results
Performance results of various models on this benchmark
| Paper Title | ||
|---|---|---|
| Unicom+ViT-L@336px | 88.9 | Unicom: Universal and Compact Representation Learning for Image Retrieval |
| EfficientDML-VPTSP-G/512 | 84.5 | Learning Semantic Proxies from Visual Prompts for Parameter-Efficient Fine-Tuning in Deep Metric Learning |
| Recall@k Surrogate loss (ViT-B/16) | 83.0 | Recall@k Surrogate Loss with Large Batches and Similarity Mixup |
| ROADMAP (DeiT-S) | 73.6 | Robust and Decomposable Average Precision for Image Retrieval |
| Recall@k Surrogate loss (ResNet-50) | 71.8 | Recall@k Surrogate Loss with Large Batches and Similarity Mixup |
| HAPPIER_F (ResNet-50) | 71.0 | Hierarchical Average Precision Training for Pertinent Image Retrieval |
| HAPPIER (ResNet-50) | 70.7 | Hierarchical Average Precision Training for Pertinent Image Retrieval |
| ROADMAP (ResNet-50) | 69.1 | Robust and Decomposable Average Precision for Image Retrieval |
| Smooth-AP | 67.2 | Smooth-AP: Smoothing the Path Towards Large-Scale Image Retrieval |
| PNP Loss | 66.6 | Rethinking the Optimization of Average Precision: Only Penalizing Negative Instances before Positive Ones is Enough |
0 of 10 row(s) selected.