Sistema electrónico de corrección del gesto deportivo en base a la biomecánica en deportistas de marcha atlética

Abstract

ABSTRACT This research project describes the design and implementation of the electronic sports gesture correction system based on biomechanics in athletic walking athletes using machine learning, deep learning and deep machine learning, these libraries and models that are aimed at this field help the implementation of the prototype for monitoring the practitioners of athletic walking through sensorization and detection, this structure is suitable for all types of practitioners, whether of different age, height or physical condition. For the operation of the processing of the collected images requires an artificial vision system that depends on convolutional neural networks (CNN) that helps with the detection of key points to draw a skeleton on the practitioner of athletic walking and proceed to the analysis of each movement. The prototype of the monitoring system consists of a presence sensor for the activation of a stepper motor that serves to move the camera support on the electronic rail in order to collect images, the support consists of servomotors that serve to position the camera at a fixed point of the athletic gait practitioner, the prototype also has limit switches that serve both to end and reset the system's operation. The acquisition of the images is to carry out the respective analysis of the movements that the marcher made both in the single support phase and the double support phase, this analysis is carried out by calculating the angles in all the stages of the phases, angles that they are sent to a database. All this procedure and thanks to the set of tests resulted in obtaining a highly efficient system for the detection and analysis of the different stages of the phases of athletic walking.