Optimized assessment of physical rehabilitation exercises using spatiotemporal, sequential graph-convolutional networks

Ikram Kourbane, Panagiotis Papadakis, Mihai Andries

Abstract: Highlights•Used GCNs to create adjacency matrices and extract temporal features.•Added action classification to ensure quality scores for valid sequences.•Achieved high performance on KIMORE and UI-PRMD with a real-time model.