Classification of physical activities based on body-segments coordination

摘要: Numerous innovations based on connected objects and physical activity (PA) monitoring have been proposed. However, recognition of PAs requires robust algorithm methodology. The current study presents an innovative approach for PA recognition. It is the heuristic definition postures use body-segments coordination obtained through external sensors. first part this methodology required to define set accelerations which most appropriate represent particular involved in chosen (here walking, running, cycling). For that purpose, subjects different ages heterogeneous conditions walked, ran, cycled, performed daily activities at paces. From 3D motion capture, vertical horizontal 8 anatomical landmarks representative body were computed. Then, 680 combinations from up 3 compared identify acceleration discriminate terms segment coordinations. discrimination was maximal Hausdorff Distance between accelerations. both knees demonstrated best discrimination. second step proof concept, implementing proposed classify new group subjects. originality possibility subject's specific measures as reference data. With algorithm, 94% trials correctly classified.In conclusion, our a flexible extendable At stage, has shown be valid subjects, suggests it could deployed clinical or health-related applications regardless subjects' abilities characteristics. An classification detailed.Body-segments by allowed classification.Knees discrimination.Proof concept with experimental data achieved.