/OpenSelfSup

Self-Supervised Learning Toolbox and Benchmark

Primary LanguagePythonApache License 2.0Apache-2.0

OpenSelfSup

News

  • A bug of MoCo v2 has been fixed and now the results are reproducible.
  • OpenSelfSup now supports BYOL!

Introduction

The master branch works with PyTorch 1.1 or higher.

OpenSelfSup is an open source unsupervised representation learning toolbox based on PyTorch.

What does this repo do?

Below is the relations among Unsupervised Learning, Self-Supervised Learning and Representation Learning. This repo focuses on the shadow area, i.e., Unsupervised Representation Learning. Self-Supervised Representation Learning is the major branch of it. Since in many cases we do not distingush between Self-Supervised Representation Learning and Unsupervised Representation Learning strictly, we still name this repo as OpenSelfSup.

Major features

  • All methods in one repository

    For comprehensive comparison in all benchmarks, refer to MODEL_ZOO.md. Most of the selfsup pretraining methods are under the batch_size=256, epochs=200 setting.

    MethodVOC07 SVM (best layer)ImageNet (best layer)
    ImageNet87.1776.17
    Random30.5416.21
    Relative-Loc64.7849.31
    Rotation-Pred67.3854.99
    DeepCluster74.2657.71
    NPID74.5056.61
    ODC78.4257.70
    MoCo79.1860.60
    MoCo v284.2667.69
    SimCLR78.9561.57
    BYOL (bs4096)85.1069.14
    • Flexibility & Extensibility

      OpenSelfSup follows a similar code architecture of MMDetection while is even more flexible than MMDetection, since OpenSelfSup integrates various self-supervised tasks including classification, joint clustering and feature learning, contrastive learning, tasks with a memory bank, etc.

      For existing methods in this repo, you only need to modify config files to adjust hyper-parameters. It is also simple to design your own methods, please refer to GETTING_STARTED.md.

    • Efficiency

      All methods support multi-machine multi-gpu distributed training.

    • Standardized Benchmarks

      We standardize the benchmarks including logistic regression, SVM / Low-shot SVM from linearly probed features, semi-supervised classification, and object detection. Below are the setting of these benchmarks.

      Benchmarks Setting Remarks
      ImageNet Linear Classification (Multi) goyal2019scaling Evaluate different layers.
      ImageNet Linear Classification (Last) MoCo Evaluate the last layer after global pooling.
      Places205 Linear Classification goyal2019scaling Evaluate different layers.
      ImageNet Semi-Sup Classification
      PASCAL VOC07 SVM goyal2019scaling Costs="1.0,10.0,100.0" to save evaluation time w/o change of results.
      PASCAL VOC07 Low-shot SVM goyal2019scaling Costs="1.0,10.0,100.0" to save evaluation time w/o change of results.
      PASCAL VOC07+12 Object Detection MoCo
      COCO17 Object Detection MoCo

    Change Log

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

    [2020-06-26] OpenSelfSup v0.2.0 is released with benchmark results and support of new features.

    [2020-06-16] OpenSelfSup v0.1.0 is released.

    Installation

    Please refer to INSTALL.md for installation and dataset preparation.

    Get Started

    Please see GETTING_STARTED.md for the basic usage of OpenSelfSup.

    Benchmark and Model Zoo

    Please refer to MODEL_ZOO.md for for a comprehensive set of pre-trained models and benchmarks.

    License

    This project is released under the Apache 2.0 license.

    Acknowledgement

    • This repo borrows the architecture design and part of the code from MMDetection.
    • The implementation of MoCo and the detection benchmark borrow the code from moco.
    • The SVM benchmark borrows the code from fair_self_supervision_benchmark.
    • openselfsup/third_party/clustering.py is borrowed from deepcluster.

    Contact

    This repo is currently maintained by Xiaohang Zhan (@XiaohangZhan) and Jiahao Xie (@Jiahao000).