OpenMixup

📘Documentation | 🛠️Installation | 🚀Model Zoo | 👀Awesome Mixup | 🔍Awesome MIM | 🆕News

Introduction

The main branch works with PyTorch 1.8 (required by some self-supervised methods) or higher (we recommend PyTorch 1.10). You can still use PyTorch 1.6 for supervised classification methods.

OpenMixup is an open-source toolbox for supervised, self-, and semi-supervised visual representation learning with mixup based on PyTorch, especially for mixup-related methods.

What's New

[2020-08-19] Weights and logs for mixup benchmarks are released.

[2020-07-30] OpenMixup v0.2.5 is released (issue #10).

Installation

There are quick installation steps for develepment:

conda create -n openmixup python=3.8 pytorch=1.10 cudatoolkit=11.3 torchvision -c pytorch -y
conda activate openmixup
pip3 install openmim
mim install mmcv-full
git clone https://github.com/Westlake-AI/openmixup.git
cd openmixup
python setup.py develop

Please refer to install.md for more detailed installation and dataset preparation.

Getting Started

Please see get_started.md for the basic usage of OpenMixup. You can start a multiple GPUs training with CONFIG_FILE using the following script. An example,

bash tools/dist_train.sh ${CONFIG_FILE} ${GPUS} [optional arguments]

Please Then, see Tutorials for more tech details:

• config files
• add new dataset
• data pipeline
• add new modules
• customize schedules
• customize runtime
• benchmarks

Overview of Model Zoo

Please refer to Model Zoos for various backbones, mixup methods, and self-supervised algorithms. We also provide the paper lists of Awesome Mixups for your reference. Checkpoints and traning logs will be updated soon!

• Backbone architectures for supervised image classification on ImageNet.

  Currently supported backbones

  • [x] VGG (ICLR'2015) [config]
  • [x] ResNet (CVPR'2016) [config]
  • [x] ResNeXt (CVPR'2017) [config]
  • [x] SE-ResNet (CVPR'2018) [config]
  • [x] SE-ResNeXt (CVPR'2018) [config]
  • [x] ShuffleNetV2 (ECCV'2018) [config]
  • [x] MobileNetV2 (CVPR'2018) [config]
  • [x] MobileNetV3 (ICCV'2019)
  • [x] EfficientNet (ICML'2019) [config]
  • [x] Swin-Transformer (ICCV'2021) [config]
  • [x] RepVGG (CVPR'2021)
  • [x] Vision-Transformer (ICLR'2021) [config]
  • [x] MLP-Mixer (NIPS'2021) [config]
  • [x] DeiT (ICML'2021) [config]
  • [x] ConvMixer (Openreview'2021) [config]
  • [x] UniFormer (ICLR'2022) [config]
  • [x] PoolFormer (CVPR'2022) [config]
  • [x] ConvNeXt (CVPR'2022) [config]
  • [x] VAN (ArXiv'2022) [config]
  • [x] LITv2 (ArXiv'2022) [config]

• Mixup methods for supervised image classification.

  Currently supported mixup methods

  • [x] Mixup (ICLR'2018)
  • [x] CutMix (ICCV'2019)
  • [x] ManifoldMix (ICML'2019)
  • [x] FMix (ArXiv'2020)
  • [x] AttentiveMix (ICASSP'2020)
  • [x] SmoothMix (CVPRW'2020)
  • [x] SaliencyMix (ICLR'2021)
  • [x] PuzzleMix (ICML'2020)
  • [x] GridMix (Pattern Recognition'2021)
  • [x] ResizeMix (ArXiv'2020)
  • [x] AutoMix (ECCV'2022) [config]
  • [x] SAMix (ArXiv'2021) [config]
  Currently supported datasets for mixups

  • [x] ImageNet [download] [config]
  • [x] CIFAR-10 [download] [config]
  • [x] CIFAR-100 [download] [config]
  • [x] Tiny-ImageNet [download] [config]
  • [x] CUB-200-2011 [download] [config]
  • [x] FGVC-Aircraft [download] [config]
  • [x] StandfoldCars [download]
  • [x] Place205 [download] [config]
  • [x] iNaturalist-2017 [download] [config]
  • [x] iNaturalist-2018 [download] [config]

• Self-supervised algorithms for visual representation.

  Currently supported self-supervised algorithms

  • [x] Relative Location (ICCV'2015) [config]
  • [x] Rotation Prediction (ICLR'2018) [config]
  • [x] DeepCluster (ECCV'2018) [config]
  • [x] NPID (CVPR'2018) [config]
  • [x] ODC (CVPR'2020) [config]
  • [x] MoCov1 (CVPR'2020) [config]
  • [x] SimCLR (ICML'2020) [config]
  • [x] MoCoV2 (ArXiv'2020) [config]
  • [x] BYOL (NIPS'2020) [config]
  • [x] SwAV (NIPS'2020) [config]
  • [x] DenseCL (CVPR'2021) [config]
  • [x] SimSiam (CVPR'2021) [config]
  • [x] Barlow Twins (ICML'2021) [config]
  • [x] MoCoV3 (ICCV'2021) [config]
  • [x] MAE (CVPR'2022) [config]
  • [x] SimMIM (CVPR'2022) [config]
  • [x] CAE (ArXiv'2022)
  • [x] A2MIM (ArXiv'2022) [config]

Change Log

Please refer to changelog.md for details and release history.

License

This project is released under the Apache 2.0 license.

Acknowledgement

• OpenMixup is an open-source project for mixup methods created by researchers in CAIRI AI LAB. We encourage researchers interested in visual representation learning and mixup methods to contribute to OpenMixup!
• This repo borrows the architecture design and part of the code from MMSelfSup and MMClassification.

Citation

If you find this project useful in your research, please consider cite our repo:

@misc{2022openmixup,
    title={{OpenMixup}: Open Mixup Toolbox and Benchmark for Visual Representation Learning},
    author={Li, Siyuan and Liu, Zichen and Wu, Di and Stan Z. Li},
    howpublished = {\url{https://github.com/Westlake-AI/openmixup}},
    year={2022}
}

Contributors

For now, the direct contributors include: Siyuan Li (@Lupin1998), Zicheng Liu (@pone7), and Di Wu (@wudi-bu). We thank contributors from MMSelfSup and MMClassification.

Contact

This repo is currently maintained by Siyuan Li ([email protected]) and Zicheng Liu ([email protected]).