# openpose_ros **Repository Path**: turing123456/openpose_ros ## Basic Information - **Project Name**: openpose_ros - **Description**: No description available - **Primary Language**: Unknown - **License**: Not specified - **Default Branch**: master - **Homepage**: None - **GVP Project**: No ## Statistics - **Stars**: 0 - **Forks**: 0 - **Created**: 2021-05-06 - **Last Updated**: 2021-05-06 ## Categories & Tags **Categories**: Uncategorized **Tags**: None ## README # openpose_ros Example ROS catkin package that utilizes the OpenPose library from https://github.com/CMU-Perceptual-Computing-Lab/openpose. ## System Tested on: * Ubuntu 14.04 / Ubuntu 16.04 * ROS Indigo / Kinetic * CUDA 8.0 / CUDA 10.0 / CUDA 10.1 * cuDNN 5.1 / cuDNN 6.0 / cuDNN 7.2.4 / cuDNN 7.5.0 * OpenCV 3.3 / OpenCV 3.4 ## Installation Steps 1. Clone OpenPose somewhere not in your catkin_workspace. ```bash git clone https://github.com/CMU-Perceptual-Computing-Lab/openpose.git ``` 2. IMPORTANT: I do not update this repository very frequently and the maintainers of OpenPose tend to change their API frequently, so I can only guarantee that it will work with certain versions of OpenPose. Currently I have updated this ros wrapper to work with commit 254570d. You can use get that version by running the following commands: ```bash cd openpose git checkout 254570d ``` 3. Install openpose using instructions from here: https://github.com/CMU-Perceptual-Computing-Lab/openpose/blob/254570df262d91b1940aaf5797ba6c5d6db4b52f/doc/installation.md. Make sure to run `sudo make install` in the build folder at the end. 4. Clone this repository into your catkin_workspace/src directory. ```bash git clone https://github.com/firephinx/openpose_ros.git ``` 5. Modify the model_folder line in openpose_ros/src/openpose_flags.cpp to where openpose is installed (line 30). ```bash DEFINE_string(model_folder, "/path/to/openpose/models/", "Folder path (absolute or relative) where the models (pose, face, ...) are located."); ``` 6. Modify the image_topic parameter in openpose_ros/launch/openpose_ros.launch to the image_topic you want to process. ```bash ``` 7. Modify the other parameters in openpose_ros/src/openpose_flags.cpp and openpose_ros/launch/openpose_ros.launch to your liking such as enabling face and hands detection. 8. Run catkin_make from your catkin_workspace directory. ### Potential Installation Issues 1. If cv_bridge is causing you errors and/or you decide to use OpenCV 3.2+, copy the cv_bridge folder from https://github.com/ros-perception/vision_opencv into your catkin_workspace/src directory. 2. If you have problems with CUDA during catkin_make, uncomment this line in CMakeLists.txt # find_package(CUDA REQUIRED) ## Running ```bash source catkin_workspace/devel/setup.bash roslaunch openpose_ros openpose_ros.launch ```