Sistema electrónico de entrenamiento de salto de longitud mediante visión artificial aplicado a deportistas con discapacidad visual

Abstract

This research project describes the development of a training system for athletes with visual impairment supported by artificial vision. Initially, a study was carried out on the current state of the difficulties that visually impaired athletes are exposed to when doing any sport, and thanks to the advancement of artificial vision in the different application areas, an electronic system based on hardware and software that allows the long jump training process to be carried out, in such a way that the athlete can be monitored and controlled, avoiding complications and injuries during training. The main objective of the electronic training system is based on the processing of images, these images depend on deep learning (Deep Learning), which reaches different levels of hierarchical details, so that the information that flows upwards is united, creating abstractions. and higher and more complex representations. For example, with digital images, pixels are combined into edges, edges into outlines, outlines into shapes, and finally shapes into objects. Obtaining as a result the recognition of the long jump track and the athletes in it. The system consists of 3 stages, the first stage of image acquisition, consisting of the sony wcx550 webcam that allows obtaining images in real time. In the second stage, processing, stable results were obtained, thanks to the use of a Raspberry Pi B3+ single-board computer, which processes the information obtained in the image acquisition stage and helps to detect people, contour detection and data management for the alert management stage. The system sends detection alerts for presence on the runway, alerts for the jump performed and stores the number of tests performed. The system will provide security and confidence, at the time of training, both to the athlete with visual impairment, and to the coach.