Recognition and Evaluation of Stability Movements in Gymnastics Based on Deep Learning

Research in the field of computer vision is currently an active and developing area. The evolution of technology and the increasing capabilities of computing and artificial neural networks are factors that have contributed to the advancement of this field. Computer vision systems are relied upon in many fields at present, such as the sports, agricultural, and medical fields, which gave excellent results in all areas on which these systems relied. This article presents the method of relying on computer vision systems in the sports field by distinguishing and evaluating the movements of the male gymnastics player on the ring apparatus, the evaluation process by the referees is complex, and the angle of view of the referee must be clear to ensure the assessment is correct, as the movements are distinguished. Stability in the fixed ring game in gymnastics, a data set was created to distinguish the activities and to evaluate the accuracy of their performance due to the lack of a data set available on the Internet, knowing that these movements are only for men’s gymnastics. One of the convolutional neural network models (yolov7) was adopted after dividing the data into 80% for training and the rest for testing. A set of points on the body based on which the type of movement is known, in addition to the player’s accuracy in executing the action. The time it takes for the player to perform this movement has also been calculated. This system has also been applied to a set of video clips to evaluate player movement. Adopting it to train some players to implement the actions well, the implementation results were very high, as the accuracy reached 95%.