Long-tail Learning
Long-tailed learning, one of the most challenging problems in visual recognition, aims to train well-performing models from a large number of images that follow a long-tailed class distribution.
Papers
Showing 1–10 of 131 papers
All datasetsImageNet-LTCIFAR-100-LT (ρ=100)CIFAR-10-LT (ρ=10)iNaturalist 2018CIFAR-100-LT (ρ=10)Places-LTCIFAR-10-LT (ρ=100)CIFAR-100-LT (ρ=50)MIMIC-CXR-LTNIH-CXR-LTCOCO-MLTVOC-MLT
Benchmark Results
| # | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| 1 | LIFT (ViT-L/14@336px) | Top-1 Accuracy | 87.4 | — | Unverified |
| 2 | LIFT (ViT-L/14) | Top-1 Accuracy | 85.2 | — | Unverified |
| 3 | GML (ViT-B-16) | Top-1 Accuracy | 82.1 | — | Unverified |
| 4 | LIFT (ViT-B/16) | Top-1 Accuracy | 80.4 | — | Unverified |
| 5 | RAC (ViT-B-16) | Top-1 Accuracy | 80.24 | — | Unverified |
| 6 | GPaCo (2-R152) | Top-1 Accuracy | 79.8 | — | Unverified |
| 7 | TADE(ResNet-152) | Top-1 Accuracy | 77 | — | Unverified |
| 8 | ProCo (ResNet50) | Top-1 Accuracy | 75.8 | — | Unverified |
| 9 | MDCS(Resnet50) | Top-1 Accuracy | 75.6 | — | Unverified |
| 10 | DeiT-LT | Top-1 Accuracy | 75.1 | — | Unverified |