andrewtal / moco-pytorch

MoCo: Momentum Contrast for Unsupervised Visual Representation Learning

Modify from: https://github.com/facebookresearch/moco

Unsupervised Training

Training

Use Multi-GPU with Pytorch:

# Training
$ python train.py --image_folder /path/to/your/images

# Ckpt convert
$ python ckpt_convert.py --ckpt_path *.ckpt --save_path *.pth --arch resnet*

To run MoCo v1, set --mlp False --moco-t 0.07 --cos False.

Citation

This is a PyTorch implementation of the MoCo paper:

@Article{he2019moco,
  author  = {Kaiming He and Haoqi Fan and Yuxin Wu and Saining Xie and Ross Girshick},
  title   = {Momentum Contrast for Unsupervised Visual Representation Learning},
  journal = {arXiv preprint arXiv:1911.05722},
  year    = {2019},
}

It also includes the implementation of the MoCo v2 paper:

@Article{chen2020mocov2,
  author  = {Xinlei Chen and Haoqi Fan and Ross Girshick and Kaiming He},
  title   = {Improved Baselines with Momentum Contrastive Learning},
  journal = {arXiv preprint arXiv:2003.04297},
  year    = {2020},
}