Lightweight Auxiliary Sports Training Quality Evaluation Based on Edge‐Cloud Collaboration

ABSTRACT With the rapid development of artificial intelligence (AI) and internet of things (IoTs) technologies, it has become a challenge task to assist sports training using intelligent system in physical education. This paper establishes a lightweight assisted sports training quality evaluation system using edge‐cloud computing and AI technology to weaken this issue. First, the skeleton of person during sports training is captured by Kinect camera; then, the skeleton is represented as joint angle feature vector; lastly, the joint angle feature vector is sent to cloud server in which an ordinal regression forest is deployed. Compared with complex deep learning model, both skeletal joint angle feature extraction and ordinal regression forest need fewer computing resources, which can adapt limited resources in devices under IoT environment. On the other hand, ordinal regression forest can reflect the ordinal relationship between sports training quality levels to reach lower mean absolute error (MAE). The experiments show the effectiveness of the proposed sports training quality evaluation system.