Research on Intelligent Algorithm-Based Swimming Athlete Pose Recognition and Correction Method

We have developed a pose recognition technology based on depth images to better assist swimming athletes in correcting their abnormal postures. This technology uses threshold algorithms for data preprocessing and Kalman filters to filter noise. We also employ Gaussian distribution functions to capture dynamic changes and utilize the SURF algorithm to remove blurry parts. This technology can help us better assist swimming athletes in correcting their abnormal postures and improving their daily training. Using the Euclidean distance method, we can accurately estimate the distance between two adjacent reference points and employ feedback monitoring techniques to correct improper postures. Through simulations, we have found that this new deep-level image skeleton tracking technology can effectively capture the dynamics of athletes and accurately detect their poses, demonstrating high accuracy and stability.