Development trend of the application of image recognition technology in the process of sports training

Abstract

In the realm of sports training, the role of accurate image recognition is increasingly crucial for the effective correction of athletic movements. This research paper delves into the application of image recognition technologies to analyze sports training actions. Initial steps include the enhancement of image quality by filtering and sharpening images captured at a sports academy. Advanced techniques such as target detection algorithms and critical frame extraction are then applied to these refined images. Evaluations conducted on the KTH and UCF Sports action datasets reveal an average recognition rate of 88%, with further breakdowns indicating lower performance in activities like walking, jogging, and fast running in the KTH dataset. In contrast, uniform recognition results are observed in the UCF dataset with an average rate of 89.2% across various actions. The findings underscore the effectiveness of image recognition in improving sports training methodologies.