liuxingwt / cls

A basic image classification framework in pytorch

Classification

This repository is an implementation of basic image classification framework.

Environment

• OS version: Ubuntu 18.04
• NVIDIA diver version: 465.27
• Cuda version: 11.3
• Python version: 3.6.9
• Python packages installation:
  pip3 install -i https://mirrors.aliyun.com/pypi/simple -r requirements.txt

Train

sh run/train.sh

Test

sh run/test.sh

Reference:

Code:

• pytroch-imageNet: https://github.com/pytorch/examples/blob/master/imagenet/main.py
• pytorch-image-models: https://github.com/rwightman/pytorch-image-models
• Vit-Pytorch: https://github.com/lucidrains/vit-pytorch
• DeiT: https://github.com/facebookresearch/deit
• Swin-Transformer: https://github.com/microsoft/Swin-Transformer

Paper:

• Vit: https://arxiv.org/abs/2010.11929
• DeiT: https://arxiv.org/abs/2012.12877
• Swin: http://arxiv.org/abs/2103.14030
• CaiT: https://arxiv.org/abs/2103.17239

Changing log:

20210524: change the order of input image channel from BGR to RGB, to better adapt to pretrained model in ImageNet.

• dataloader.py：修改了图片加载顺序BGR->RGB
• test.py：修改了图片加载顺序BGR->RGB
• train.py: 在一个epoch内保存模型时，只保存权重

20211203:

• optim/: 增加了ranger优化器，并修改了build_optimizer.py和train.py文件
• tools/metrics.py：重构了二分类结果evaluation的流程代码