OpenPose

资源分类

OpenPose

2019-08-19 |

1192 |

0 |

OpenPose

| Linux | |-------------| ||

OpenPose represents the first real-time multi-person system to jointly detect human body, hand, and facial keypoints (in total 130 keypoints) on single images.

Features

Functionality:

Easy estimation of distortion, intrinsic, and extrinsic camera parameters.
3-D triangulation from multiple single views.
Synchronization of Flir cameras handled.
Compatible with Flir/Point Grey cameras, but provided C++ demos to add your custom input.
15 or 18 or 25-keypoint body/foot keypoint estimation. Running time invariant to number of detected people.
2x21-keypoint hand keypoint estimation. Currently, running time depends on number of detected people.
70-keypoint face keypoint estimation. Currently, running time depends on number of detected people.
2D real-time multi-person keypoint detection:
3D real-time multi-person keypoint detection:
Calibration toolbox:

Input: Image, video, webcam, Flir/Point Grey and IP camera. Included C++ demos to add your custom input.
Output: Basic image + keypoint display/saving (PNG, JPG, AVI, ...), keypoint saving (JSON, XML, YML, ...), and/or keypoints as array class.
OS: Ubuntu (14, 16), Windows (8, 10), Mac OSX, Nvidia TX2.
Others:

Available: command-line demo, C++ wrapper, and C++ API.
CUDA (Nvidia GPU) and CPU versions.

Latest Features

Jun 2018: Combined body-foot model released! Faster and more accurate.!
Jun 2018: Python version released!
Jun 2018: AMD graphic card version released!
Jun 2018: Calibration toolbox released!
Jun 2018: Mac OSX version (CPU) released!
Mar 2018: CPU version!
Mar 2018: 3-D keypoint reconstruction module (from multiple camera views)!
Sep 2017: CMake installer and IP camera support!
Jul 2017: Windows portable binaries and demo!
Jul 2017: Hands released!
Jun 2017: Face released!

For further details, check all released features and release notes.

Results

Body Estimation

3-D Reconstruction Module

Body, Face, and Hands Estimation

Unity Plugin

Runtime Analysis

Inference time comparison between the 3 available pose estimation libraries: OpenPose, Alpha-Pose (fast Pytorch version), and Mask R-CNN:

This analysis was performed using the same images for each algorithm and a batch size of 1. Each analysis was repeated 1000 times and then averaged. This was all performed on a system with a Nvidia 1080 Ti and CUDA 8. Megvii (Face++) and MSRA GitHub repositories were excluded because they only provide pose estimation results given a cropped person. However, they suffer the same problem than Alpha-Pose and Mask R-CNN, their runtimes grow linearly with the number of people.

Installation, Reinstallation and Uninstallation

Windows portable version: Simply download and use the latest version from the Releases section.

Otherwise, check doc/installation.md for instructions on how to build OpenPose from source.

Quick Start

Most users do not need the OpenPose C++ API, but they can simply use the basic Demo and/or OpenPose Wrapper.

Demo: To easily process images/video/webcam and display/save the results. See doc/demo_overview.md. E.g. run OpenPose in a video with:

# Ubuntu
./build/examples/openpose/openpose.bin --video examples/media/video.avi
:: Windows - Portable Demo
binOpenPoseDemo.exe --video examplesmediavideo.avi

Calibration toolbox: To easily calibrate your cameras for 3-D OpenPose or any other stereo vision task. See doc/calibration_demo.md.
OpenPose Wrapper: If you want to read a specific input, and/or add your custom post-processing function, and/or implement your own display/saving, check the Wrapper tutorial on examples/tutorial_wrapper/. You can create your custom code on examples/user_code/ and quickly compile it by using make all in the OpenPose folder (assuming Makefile installer).
OpenPose C++ API: See doc/library_introduction.md.
Adding an extra module: Check doc/library_add_new_module.md.
Standalone face or hand detector:

Face keypoint detection without body keypoint detection: If you want to speed it up (but also reduce amount of detected faces), check the OpenCV-face-detector approach in doc/standalone_face_or_hand_keypoint_detector.md.
Use your own face/hand detector: You can use the 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 (OpenPose detector would fail). See doc/standalone_face_or_hand_keypoint_detector.md.

Library dependencies: OpenPose uses default Caffe and OpenCV, as well as any Caffe dependency. The demos additionally use GFlags. It could easily be ported to other deep learning frameworks (Tensorflow, Torch, ...). Feel free to make a pull request if you implement any of those!

Output

Output (format, keypoint index ordering, etc.) in doc/output.md.

Speeding Up OpenPose and Benchmark

Check the OpenPose Benchmark as well as some hints to speed up and/or reduce the memory requirements for OpenPose on doc/faq.md#speed-up-memory-reduction-and-benchmark.

Send Us Failure Cases and Feedback!

Our library is open source for research purposes, and we want to continuously improve it! So please, let us know 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!
... 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!

Authors and Contributors

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 Yaadhav Raaj. The original CVPR 2017 repo includes Matlab and Python versions, as well as the training code. The body pose estimation work is based on the original ECCV 2016 demo.

In addition, OpenPose would not be possible without the CMU Panoptic Studio dataset.

We would also like to thank all the people who helped OpenPose in any way. The main contributors are listed in doc/contributors.md.

Citation

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}
}

License

OpenPose is freely available for free non-commercial use, and may be redistributed under these conditions. Please, see the license for further details. Interested in a commercial license? Check this link. For commercial queries, contact Yaser Sheikh.

下一篇：Mask R-CNN

用户评价