OpenPose is a library for real-time multi-person keypoint detection and multi-threading written in C++ using OpenCV and Caffe.
- Sep 2017: CMake installer and IP camera support!
- Jul 2017: Windows portable demo!
- Jul 2017: Hands released!
- Jun 2017: Face released!
- May 2017: Windows version!
- Apr 2017: Body released!
- Check all the release notes.
- Latest News
- Results
- Introduction
- Functionality
- Installation, Reinstallation and Uninstallation
- Quick Start
- Output
- Standalone Face Or Hand Keypoint Detector
- Speed Up Openpose And Benchmark
- Send Us Failure Cases!
- Send Us Your Feedback!
- Citation
- Other Contributors
OpenPose represents the first real-time system to jointly detect human body, hand and facial keypoints (in total 130 keypoints) on single images. In addition, the system computational performance on body keypoint estimation is invariant to the number of detected people in the image. It uses Caffe, but it could easily be ported to other frameworks (Tensorflow, Torch, etc.). If you implement any of those, feel free to make a pull request!
OpenPose is authored by Gines Hidalgo, Zhe Cao, Tomas Simon, Shih-En Wei, Hanbyul Joo, and Yaser Sheikh. Currently, it is being maintained by Gines Hidalgo and Bikramjot Hanzra.
It is freely available for free non-commercial use, and may be redistributed under these conditions. Please note, that all derivative work will become property of Carnegie Mellon University and will be licensed under the same conditions to the contributing author. Please, see the license for further details. Interested in a commercial license? Check this link. For commercial queries, contact Yaser Sheikh.
In addition, OpenPose would not be possible without the CMU Panoptic Studio.
The pose estimation work is based on the C++ code from the ECCV 2016 demo, "Realtime Multiperson Pose Estimation", Zhe Cao, Tomas Simon, Shih-En Wei, Yaser Sheikh. The original repo includes Matlab and Python version, as well as the training code.
- Multi-person 15 or 18-keypoint body pose estimation and rendering. Running time invariant to number of people on the image.
- Multi-person 2x21-keypoint hand estimation and rendering. Note: In this initial version, running time linearly depends on the number of people on the image.
- Multi-person 70-keypoint face estimation and rendering. Note: In this initial version, running time linearly depends on the number of people on the image.
- Flexible and easy-to-configure multi-threading module.
- Image, video, webcam and IP camera reader.
- Able to save and load the results in various formats (JSON, XML, PNG, JPG, ...).
- Small display and GUI for simple result visualization.
- All the functionality is wrapped into a simple-to-use OpenPose Wrapper class.
You can find the installation, reinstallation and uninstallation steps on: doc/installation.md.
Most users do not need the OpenPose C++ API, but they can simply use the basic Demo and/or OpenPose Wrapper.
Ideal to process images/video/webcam and display/save the results. Check doc/demo_overview.md.
Ideal if you want to read a specific input, and/or add your custom post-processing function, and/or implement your own display/saving. Take a look to the Wrapper tutorial on examples/tutorial_wrapper/. You might create your custom code on examples/user_code/ and compile it by using make all in the OpenPose folder.
Learn how to easily add an extra module to OpenPose in doc/library_add_new_module.md.
Your case if you want to use the C++ API. See doc/library_introduction.md.
Output (format, keypoint index ordering, etc.) in doc/output.md.
If you do not need the body detector and want to speed up the face keypoint detection, you can use the OpenCV-based approach, see doc/standalone_face_or_hand_keypoint_detector.md.
You can also use the OpenPose hand and/or face keypoint detectors with your own face or hand detectors, rather than using the body detector. E.g. useful for camera views at which the hands are visible but not the body, so that the OpenPose detector would fail. See doc/standalone_face_or_hand_keypoint_detector.md.
Check the OpenPose Benchmark and some hints to speed up OpenPose on doc/installation.md#faq.
If you find videos or images where OpenPose does not seems to work well, feel free to send them to openposecmu@gmail.com (email only for failure cases!), we will use them to improve the quality of the algorithm. Thanks!
Our library is open source for research purposes, and we want to continuously improve it! So please, let us know if...
-
... you find any bug (in functionality or speed).
-
... you added some functionality to some class or some new Worker subclass which we might potentially incorporate.
-
... you know how to speed up or improve any part of the library.
-
... you have a request about possible functionality.
-
... etc.
Just comment on GitHub or make a pull request and we will answer as soon as possible! Send us an email if you use the library to make a cool demo or YouTube video!
Please cite these papers in your publications if it helps your research (the face keypoint detector was trained using the same procedure described in [Simon et al. 2017]):
@inproceedings{cao2017realtime,
author = {Zhe Cao and Tomas Simon and Shih-En Wei and Yaser Sheikh},
booktitle = {CVPR},
title = {Realtime Multi-Person 2D Pose Estimation using Part Affinity Fields},
year = {2017}
}
@inproceedings{simon2017hand,
author = {Tomas Simon and Hanbyul Joo and Iain Matthews and Yaser Sheikh},
booktitle = {CVPR},
title = {Hand Keypoint Detection in Single Images using Multiview Bootstrapping},
year = {2017}
}
@inproceedings{wei2016cpm,
author = {Shih-En Wei and Varun Ramakrishna and Takeo Kanade and Yaser Sheikh},
booktitle = {CVPR},
title = {Convolutional pose machines},
year = {2016}
}
We would like to thank all the people who helped OpenPose in any way. The main contributors are listed in doc/contributors.md.