Introduction

MMFlow is a PyTorch-based optical flow toolbox and is a member of the OpenMMLab project.

The master branch code currently supports PyTorch 1.5 and aboveversion of.

main features

• The First Unified Framework for Optical Flow Algorithms

  MMFlow is the first toolbox to provide a unified implementation and evaluation framework for optical flow methods.

• modular design

  MMFlow decouples the optical flow estimation framework into different module components. By combining different module components, users can easily build a custom optical flow algorithm model.

• Rich out-of-the-box algorithms and datasets

  MMFlow supports many mainstream classic optical flow algorithms, such as FlowNet, PWC-Net, RAFT, etc., as well as the preparation and construction of various data sets, such as FlyingChairs, FlyingThings3D, Sintel, KITTI, etc.

update log

The latest v0.5.1 version has been released on 2022.07.29:

• Set the highest version of MMCV less than 1.7.0
• Update the qq_group_qrcode image in resources

If you want to know more version update details and historical information, please refer to the update log.

Install

Please refer to the installation documentation for installation, and refer to Data Preparation to prepare the dataset.

quick start

If you are new to the optical flow algorithm, you can start to understand the basic concepts of optical flow and the framework of MMFlow from learn the basics. If you are familiar with optical flow, please refer to getting_started to get started with MMFlow.

MMFlow also provides other more detailed tutorials, including:

Benchmarks and Model Libraries

Test results and models can be found in the Model Library.

Already supported algorithms:

Contribution Guidelines

We thank all contributors for their efforts to improve and enhance MMFlow. Please refer to the Contributing Guidelines for guidelines on contributing to the project.

quote

If you find this project useful for your research, please consider citing:

@misc{2021mmflow,
    title={{MMFlow}: OpenMMLab Optical Flow Toolbox and Benchmark},
    author={MMFlow Contributors},
    howpublished = {\url{https://github.com/open-mmlab/mmflow}},
    year={2021}
}

open source license

The project uses the Apache 2.0 open source license.

Other OpenMMLab projects

• MMCV: OpenMMLab Computer Vision Basic Library

• MIM: MIM is a unified entry for OpenMMlab projects, algorithms, and models

• MMClassification: OpenMMLab Image Classification Toolbox

• MMDetection: OpenMMLab Object Detection Toolbox

• MMDetection3D: OpenMMLab’s next-generation general-purpose 3D object detection platform

• MMRotate: OpenMMLab Rotating Box Detection Toolbox and Benchmark

• MMSegmentation: OpenMMLab Semantic Segmentation Toolbox

• MMOCR: OpenMMLab full-process text detection, recognition and understanding toolkit

• MMPose: OpenMMLab Pose Estimation Toolbox

• MMHuman3D: OpenMMLab Human Parametric Modeling Toolbox and Benchmark

• MMSelfSup: OpenMMLab self-supervised learning toolbox and benchmark

• MMRazor: OpenMMLab Model Compression Toolbox and Benchmark

• MMFewShot: OpenMMLab few-shot learning toolbox and benchmark

• MMAction2: OpenMMLab Next Generation Video Understanding Toolbox

• MMTracking: OpenMMLab all-in-one video object perception platform

• MMFlow: OpenMMLab Optical Flow Estimation Toolbox and Benchmark

• MMEditing: OpenMMLab image and video editing toolbox

• MMGeneration: OpenMMLab image and video generation model toolbox

• MMDeploy: OpenMMLab Model Deployment Framework

Welcome to the OpenMMLab community

Scan the QR code below to follow the official Zhihu account of the OpenMMLab team, join the official communication QQ group of the OpenMMLab team or contact the official WeChat assistant of OpenMMLab

We will serve you in the OpenMMLab community

• 📢 Share the cutting-edge core technology of AI framework
• 💻 Interpretation of PyTorch common module source code
• 📰 Release news about OpenMMLab
• 🚀 Introducing cutting-edge algorithms developed by OpenMMLab
• 🏃 Get more efficient question answering and feedback
• 🔥 Provide a platform for full communication with developers from all walks of life

Full of dry goods 📘, waiting for you to tease 💗, the OpenMMLab community is looking forward to your joining 👬