RAMS-Trans: Recurrent Attention Multi-scale Transformer forFine-grained Image Recognition

Benchmark Model Rank Results
fine-grained-image-classification-on-stanford-dogsRAMS-Trans–Accuracy: 92.4%