Image Classification on CIFAR100 LT 1.0 (test)
51.7Top-1 Acc (All)GLAG
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| GLAGTraining strategy=Data augmentation2022.02 | 51.7 | — | — | — | |
| Balanced SoftmaxTraining strategy=End-to-end training2022.02 | 50.8 | — | — | — | |
| ACEType=One-Stage, Multi-experts=42021.08 | 49.6 | 66.3 | 52.8 | 27.2 | |
| ACEType=One-Stage, Multi-experts=32021.08 | 49.4 | 66.1 | 55.7 | 23.5 | |
| RIDEType=Multi-Stage, Multi-experts=42021.08 | 49.1 | 69.3 | 49.3 | 26 | |
| MetaSAug-LDAMTraining strategy=Data augmentation2022.02 | 48.1 | — | — | — | |
| RIDEType=Multi-Stage, Multi-experts=32021.08 | 48 | 68.1 | 49.2 | 23.9 | |
| CAMType=Multi-Stage, Multi-experts=true2021.08 | 47.8 | — | — | — | |
| Bag of tricksTraining strategy=Decoupled training2022.02 | 47.8 | — | — | — | |
| RIDEType=Multi-Stage, Multi-experts=22021.08 | 47 | 67.9 | 48.4 | 21.8 | |
| MISLASTraining strategy=Decoupled training2022.02 | 47 | — | — | — | |
| Remix-DRWTraining strategy=Data augmentation2022.02 | 46.8 | — | — | — | |
| SSDTraining strategy=Data augmentation2022.02 | 46 | — | — | — | |
| RSGTraining strategy=Data augmentation2022.02 | 44.6 | — | — | — | |
| Logit AdjustmentType=One-Stage2021.08 | 43.9 | — | — | — | |
| Logit AdjustTraining strategy=End-to-end training2022.02 | 43.89 | — | — | — | |
| LDAM+LFMEType=Multi-Stage, Multi-experts=true2021.08 | 43.8 | — | — | — | |
| LDAM+M2mType=Multi-Stage2021.08 | 43.5 | — | — | — | |
| M2mTraining strategy=Data augmentation2022.02 | 43.5 | — | — | — | |
| cRTType=Multi-Stage2021.08 | 43.3 | 64 | 44.8 | 18.1 | |
| DecoupleTraining strategy=Decoupled training2022.02 | 43.3 | — | — | — | |
| τ-normType=Multi-Stage2021.08 | 43.2 | 65.7 | 43.6 | 17.3 | |
| BBNTraining strategy=Decoupled training2022.02 | 42.6 | — | — | — | |
| LDAM+DRWType=Multi-Stage2021.08 | 42 | 61.5 | 41.7 | 20.2 | |
| LDAM-DRWTraining strategy=End-to-end training2022.02 | 42 | — | — | — | |
| MixupType=One-Stage2021.08 | 41.2 | 70.7 | 40.4 | 8.8 | |
| ReMixType=One-Stage2021.08 | 40.9 | 69.6 | 40.7 | 8.8 | |
| Cross-EntropyTraining strategy=End-to-end training2022.02 | 39.8 | — | — | — | |
| ProgressiveType=One-Stage2021.08 | 39.4 | 63.3 | 38.8 | 13.1 | |
| BBNType=One-Stage, Multi-experts=true2021.08 | 39.4 | 47.2 | 49.4 | 19.8 | |
| CB Focal lossType=One-Stage, Class-balanced=true2021.08 | 38.7 | 65 | 37.6 | 10.3 | |
| Focal LossTraining strategy=End-to-end training2022.02 | 38.4 | — | — | — | |
| BaselineType=One-Stage, Backbone=ResNet-322021.08 | 38.3 | 65.2 | 37.1 | 9.1 | |
| Focal lossType=One-Stage2021.08 | 37.4 | 64.3 | 37.4 | 7.1 | |
| CB resamplingType=One-Stage, Class-balanced=true2021.08 | 36 | 59 | 35.4 | 10.9 |