Quick question regarding Kinect gesture recognition. I am looking into using the Gesture Recognition Toolkit by Nick Gillian to detect still poses (not movement) and am wondering about different skeleton sizes (kids and adults).
In order to take the different skeleton sizes into account, do I need to train the model on joint angles or will the positions suffice? Another option I thought about was to scale all skeletons to the same size before trying to detect a gesture. Has anyone solved this issue before?