Abstract

Health professionals often prescribe patients to perform specific exercises for rehabilitation of several diseases (e.g., stroke, Parkinson, backpain). When patients perform those exercises in the absence of an expert (e.g., physicians/therapists), they cannot assess the correctness of the performance. Automatic assessmentof physicalrehabilitation exercises aims to assign a quality score given an RGBD video of the body movement as input. Recent deep learning approaches address this problem by extracting CNN features from co-ordinate grids of skeleton data(body-joints) obtained from videos. However, they could not extract rich spatio-temporal features from variable-length inputs. To address this issue, we investigate Graph Convolutional Networks (GCNs) for this task. We adapt spatiotemporal GCN to predict continuous scores(assessment) instead of discrete class labels. Our model can process variable-length inputs so that users can perform any number of repetitions of the prescribed exercise. Moreover, our novel design also provides self-attention of bodyjoints, indicating their role in predicting assessment scores. It guides the user to achieve a better score in future trials by matching the same attention weights of expert users. Our model successfully outperforms existing exercise assessment methods on KIMORE and UI-PRMD datasets

Overall Results

Below you will find quantitative results for exercise assement in comparison with the previous methods.

Below you will find visualization of attention maps (red circles) and the role of body joints for five exercises. The larger circle represents the higher role of that joints. (a) Average attention map and joint role of expert users. (b) and (c) left: Role of joints when patients score high and low respectively, right: the role difference from the expert representing where (violet circles) to emphasize to get better assessment score.

Citation