An IoT Monitoring System Based on Artificial Intelligence Image Recognition and EMG Signal Processing for Abdominal Exercise Performance

• Autores: Jurado, Marco, Palma, Branco, Figueroa, Andres, Kemper, Guillermo
• Formato: artículo
• Fecha de Publicación: 2025
• Institución: Universidad Peruana de Ciencias Aplicadas
• Repositorio: UPC-Institucional
• Lenguaje: inglés
• OAI Identifier: oai:repositorioacademico.upc.edu.pe:10757/684685
• Enlace del recurso: http://hdl.handle.net/10757/684685
• Nivel de acceso: acceso abierto
• Materia: abdominal exercise; biomechanical monitoring; electromyographic (EMG) sensors; force measurement; image processing

Correctly executing exercises during training is of vital importance to ensure adequate athletic performance. Sit-ups are among the most frequently performed exercises requiring proper evaluation. This exercise contributes to increasing abdomen strength, having better posture to reduce back problems, and improving overall physical condition and appearance, among other benefits. Existing methods for evaluating the correct execution of sit-ups are manual, subjective, and inefficient in terms of time, cost, and precision. Therefore, there is a need to have technological tools that measure and monitor core abdominal strength while simultaneously verifying, through image processing, the correct execution of the exercise. Since no solutions with these capabilities have been found in the literature, this work proposes a system that performs these functions using electromyographic (EMG) sensors, force signal processing, and biomechanical monitoring based on image processing and the BlazePose algorithm. The results obtained show a very satisfactory performance of the biomechanical monitoring method, where an accuracy of over 95% is obtained in the identification of the correct body posture, while for the estimation of abdominal strength, a sensitivity of over 90% is achieved during the execution of sit-ups.