Kinect IR Tracking - Side Skeleton

Accurate tracking of the side body using the Kinect IR camera with IR reflective tapes.

The software was developed for a specific use case which is to detect if a cyclist's position on its bike is optimal or not (bike fitting) - by calculating various angles such as knee flexion, trunk angle, shoulder angle and elbow angle.

A piece of reflective tape is placed on different joints of the cyclist's body in order to let the software create and track the side skeleton. This solution was used as the skeleton tracking provided by the Kinect SDK wasn't accurate and precise enough.

The software can work for both a road bike fit test as well as a triathlon bike fit test. It automatically detects the skeleton when the users is sitted on its bike.

Made with openframeworks 0.9.8

Compilation

If you want to compile the project by yourself, you would need to do the following:

• Download openframeworks 0.9.8
• Install VS Code 2015
• Clone this repository into the openframeworks/apps/myApps folder
• Clone ofxCv into the openframeworks/addons folder. NOTE: After cloning, you have to do a git checkout stable to have the code that works with OF 0.9.8
• Clone ofxKinectForWindows2 into the openframeworks/addons folder. NOTE: After cloning you have to do a git checkout tags/0.9.0 to have the code that works with OF 0.9.8

Usage

• 'r' : Record / Stop record a video (saved in /bin/data/sessions/)
• 'g' : Show / Hide settings
• '1' : IR image
• '2' : Color image
• 'Space' : Reset skeleton
• 'a' : Manually construct the skeleton by putting the mouse pointer on each blob and pressing 'a'. In the following order : wrist -> elbow -> shoulder -> ankle -> knee -> foot

Dependencies

Install Kinect SDK 2.0 for Windows Install ffmpeg and add ffmpeg to user PATH variable

OF Addons

• ofxGui
• ofxKinectForWindows2
• ofxCv
• ofxOpenCv
• ofxVideoRecorder (modified use the one in addons/ folder)