MetaSAug: Meta Semantic Augmentation for Long-Tailed Visual Recognition

Open paper
Benchmark Model Rank Results
image-classification-on-inaturalistMetaSAug#16Top 1 Accuracy: 63.28%
image-classification-on-inaturalist-2018MetaSAug#37Top-1 Accuracy: 68.75%
long-tail-learning-on-cifar-10-lt-r-10MetaSAug-LDAM#20Error Rate: 10.32
long-tail-learning-on-cifar-10-lt-r-100MetaSAug-LDAM#18Error Rate: 19.34
long-tail-learning-on-cifar-10-lt-r-50MetaSAug-LDAM#8Error Rate: 15.66
long-tail-learning-on-cifar-100-lt-r-10MetaSAug-LDAM#18Error Rate: 38.72
long-tail-learning-on-cifar-100-lt-r-100MetaSAug-LDAM#32Error Rate: 51.99
long-tail-learning-on-cifar-100-lt-r-50MetaSAug-LDAM#21Error Rate: 47.73
long-tail-learning-on-imagenet-ltMetaSAug (ResNet-152)#52Top-1 Accuracy: 50.03
long-tail-learning-on-imagenet-ltMetaSAug with CE loss#53Top-1 Accuracy: 47.39
long-tail-learning-on-inaturalist-2018MetaSAug#36Top-1 Accuracy: 68.75%

Rank counts only results with a code link.