English | 简体中文

Introduction

MMSelfSup is an open source unsupervised representation learning toolbox based on PyTorch. It is a part of the OpenMMLab project.

The master branch works with PyTorch 1.5 or higher.

Major features

• Methods in one repository

  MMSelfsup provides state-of-the-art methods in self-supervised learning. For comprehensive comparison in all benchmarks, most of the pretraining methods are under the same setting.

• Modular Design

  MMSelfSup follows a similar code architecture of OpenMMLab projects with modular design, which is flexible and convenient for users to build their own algorithms.

• Standardized Benchmarks

  MMSelfSup standardizes the benchmarks including logistic regression, SVM / Low-shot SVM from linearly probed features, semi-supervised classification, object detection and semantic segmentation.

License

This project is released under the Apache 2.0 license.

ChangeLog

MMSelfSup v0.4.0 was released with refactor in 13/12/2021.

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

A comparison between MMSelfSup and OpenSelfSup codebases can be found in compatibility.md.

Benchmark and Model Zoo

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

Supported algorithms:

• Relative Location (ICCV'2015)
• Rotation Prediction (ICLR'2018)
• DeepCLuster (ECCV'2018)
• NPID (CVPR'2018)
• ODC (CVPR'2020)
• MoCo v1 (CVPR'2020)
• SimCLR (ICML'2020)
• MoCo v2 (ArXiv'2020)
• BYOL (NeurIPS'2020)
• SwAV (NeurIPS'2020)
• DenseCL (CVPR'2021)
• SimSiam (CVPR'2021)

More algorithms are in our plan.

Benchmarks	Setting	Remarks
ImageNet Linear Classification (Multi-head)	Goyal2019	Evaluate different layers.
ImageNet Linear Classification		Evaluate the last layer after global pooling, integrate several settings from different papers
ImageNet Semi-Sup Classification
Places205 Linear Classification (Multi-head)	Goyal2019	Evaluate different layers.
iNaturalist 2018 Classification	MoCo	End-to-End Fine-tune
PASCAL VOC07 SVM	Goyal2019	Costs="1.0,10.0,100.0" to save evaluation time w/o change of results.
PASCAL VOC07 Low-shot SVM	Goyal2019	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
Cityscapes Segmentation	MMSeg
PASCAL VOC12 Aug Segmentation	MMSeg

Installation

Please refer to install.md for installation and data_prepare.md for dataset preparation.

Get Started

Please see getting_started.md for the basic usage of MMSelfSup.

We also provides tutorials for more details:

• config
• add new dataset
• data pipeline
• add new module
• customize schedules
• customize runtime
• benchmarks

Citation

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

@misc{mmselfsup2021,
    title={OpenMMLab's Unsupervised Representation Learning Toolbox and Benchmark},
    author={MMSelfSup Contributors},
    howpublished={\url{https://github.com/open-mmlab/mmselfsup}},
    year={2021}
}

Contributing

We appreciate all contributions improving MMSelfSup. Please refer to CONTRIBUTING.md more details about the contributing guideline.

Acknowledgement

• The implementation of MoCo and the detection benchmark borrow the code from MoCo.
• The implementation of SwAV borrow the code from SwAV.
• The SVM benchmark borrows the code from fair_self_supervision_benchmark.
• mmselfsup/utils/clustering.py is borrowed from deepcluster.

Projects in OpenMMLab

• MMCV: OpenMMLab foundational library for computer vision.
• MIM: MIM Installs OpenMMLab Packages.
• MMClassification: OpenMMLab image classification toolbox and benchmark.
• MMDetection: OpenMMLab detection toolbox and benchmark.
• MMDetection3D: OpenMMLab's next-generation platform for general 3D object detection.
• MMSegmentation: OpenMMLab semantic segmentation toolbox and benchmark.
• MMAction2: OpenMMLab's next-generation action understanding toolbox and benchmark.
• MMTracking: OpenMMLab video perception toolbox and benchmark.
• MMPose: OpenMMLab pose estimation toolbox and benchmark.
• MMEditing: OpenMMLab image and video editing toolbox.
• MMOCR: OpenMMLab toolbox for text detection, recognition and understanding.
• MMGeneration: OpenMMlab toolkit for generative models.
• MMFlow: OpenMMLab optical flow toolbox and benchmark.
• MMFewShot: OpenMMLab few shot learning toolbox and benchmark.
• MMHuman3D: OpenMMLab 3D human parametric model toolbox and benchmark.