Computer vision feedback in physical education: Effects on motor performance, motivation, and movement error detection in long jump learning

Abstract

Although advances in artificial intelligence (AI)-assisted feedback technology are spreading worldwide, empirical research on the effects of feedback on the motor learning process in educational contexts remains limited. Even fewer empirical studies have been performed in physical education and sports, where learners rely heavily on feedback from teachers and equipment to acquire psychomotor skills. This quasi-experimental study examines the effect of computer vision-based feedback on long jump learning among 87 Moroccan high school students. All participants were approximately 16.5 years old, and 53% were female. The experiment took place during physical education classes. The objective was to assess the impact of three forms of feedback (AI-based feedback delivered through computer vision technology, video feedback, and oral feedback) on performance, motivation, and error perception. Accordingly, 87 high school students were divided into three groups of 25, 29, and 33 learners, respectively, referred to as Groups 1, 2, and 3. Each group underwent a diagnostic test before four trials and, finally, a concluding evaluation test at the end of the experiment. Two psychological questionnaires were used to monitor the students' psychological progress in the experiment: the Exercise Motivation Scale-28 (EMS-28) and the Movement Specific Reinvestment Scale (MSRS), including the factors of conscious movement processing (CMP) and movement self-consciousness (MSC). Data analysis revealed highly significant differences among the three groups. Learners in Group 1 improved by approximately 21.5%. Their performance scores were significantly higher than those recorded for Groups 2 and 3 by approximately 9% and 7%, respectively. Learners using computer vision technology (Group 1) achieved the highest percentages on five motivational aspects and showed deeper insight into their mistakes compared with the other two groups. The scores for the CMP and MSC factors among learners in Group 1 also increased remarkably by approximately 16.7% and 30.8%, respectively, compared with the other two groups. The scores of the latter two groups remained almost unchanged. It can also be seen from the PCA plot (Figure 2) that there are high correlations among the variables (with r-values greater than 0.70) in Group 1. These results underscore the effectiveness of computer vision technology in improving the learning of physical education skills. The findings confirm that AIassisted computer vision feedback is a powerful pedagogical tool in physical education, significantly improving motor performance, motivation, and error perception. This approach shows clear advantages over traditional feedback methods and emphasizes its strong potential for integration into modern educational practices.