This repository aims to recognize gestures made by users. Recognized gestures can lead the robot to take certain actions and/or repeat user's movement.

• Compatible Platforms
• Hardware Requirements
• Software Prerequisities
• Installation
• License

This repository is written in standards-conforming Python. It is developed and tested on the following platforms:

1. Ubuntu 16.04
2. ROS Kinetic
3. Openpose Software
4. CUDA 8.0
5. OpenCV 3.2+
6. Librealsense SDK 2.10.0+

Project uses following hardware:

1. Intel Core-i7 6700HQ
2. Nvidia Mobile Gtx 960
3. RealSense SR300 (Firmware 3.10.10.0+)
4. USB 3.0 Type-A port for camera connection

As the ROS will be main environment it needs to be installed first. You may switch between camera and openpose installation. However completing step 3 before 4 is vital to prevent kernel errors.

1. Install ROS Kinetic.

   http://wiki.ros.org/kinetic/Installation/Ubuntu

2. Install Openpose and OpenposeROS node following instructions below.

   https://github.com/firephinx/openpose_ros

Unless you want to use a separate camera skip step 3 and 4.

3. Install Realsense Camera Prerequisities.

   http://wiki.ros.org/librealsense#Installation_Prerequisites

4. Install Realsense Camera ROS Node from source.

   http://wiki.ros.org/realsense_camera/Tutorials/Building_librealsense_from_Sources

1. Git clone the package into catkin workspace src folder.

   cd ~/catkin_ws git clone https://github.com/tolgasaglik/Gesture-Recognition-with-Openpose-ROS.git

2. Go into the project folder and initialize your workspace.

   cd src/gesture_detector catkin_init_workspace

3. Build your package.

   cd ~/catkin_ws catkin_make

4. Source you environment.

   source devel/setup.bash

5. Invoke the launcher. You may modify it if you want to use another camera.

   roslaunch $(find gesture_detector)/launch/gesture_detector.launch

Unless otherwise noted, gesture_detector repository is licensed under a BSD license (see LICENSE.bsd).

This is a non-exhaustive list of contributors:

• Dr. Francisco Lera