Segmenting ARAT exercise videos to assist with the rating of the clinicians
Labeling the segments correctly to enable machine learning models to learn different movement patterns
3D reconstruction of body pose from 2D images using stereo camera setup
3D reconstruction of the hand to better understand the hand -object interaction
Generating recommendations for the clinicians based on the automated movement quality assessment
Training the models to learn the HBM weights to predict probability for a component to influence functionality