图像分类tricks

1. framework

pytorchimageclassification: https://github.com/hysts/pytorch_image_classification

2. data augmentation

auto-augment：https://github.com/DeepVoltaire/AutoAugment/blob/master/autoaugment.py
fast-autoaugment: https://github.com/kakaobrain/fast-autoaugment
augmix: https://github.com/google-research/augmix
mixup/cutout: https://github.com/PistonY/torch-toolbox
mixup/cutmix is all you need: https://www.kaggle.com/c/bengaliai-cv19/discussion/126504
Gridmask augmentation: https://www.kaggle.com/haqishen/gridmask

3. pretrained models

pretrained-models.pytorch: https://github.com/Cadene/pretrained-models.pytorch

4. metrick learning

pytorch-metric-learning: https://github.com/KevinMusgrave/pytorch-metric-learning
topkoptimization: https://github.com/BG2CRW/top_k_optimization

5. loss

Class-balanced-loss-pytorch: https://github.com/vandit15/Class-balanced-loss-pytorch
OHEM loss implementation: https://www.kaggle.com/c/bengaliai-cv19/discussion/128637
Focal loss + OHEM implementation: https://www.kaggle.com/c/bengaliai-cv19/discussion/128665

6. label smothing

mixup/cutmix with label smoothing: https://www.kaggle.com/c/bengaliai-cv19/discussion/128115