Objects detection toward complicated high remote basketball sports by leveraging deep CNN architecture

Abstract

The analysis of high-difficulty action recognition technology in basketball is mainly to identify and analyze the physical behavior of basketball players in the video to complete the technical action. The purpose of video recognition is to provide an important guarantee for improving the level of basketball training. The current target recognition technology has achieved some results. It shows that the application of target detection technology in basketball sports scene is of great significance and can improve the effect of sports training. However, traditional sports target recognition is limited by technology and injury, and the analysis of difficult sports skills is limited by the scene, dynamic background and technology, and cannot achieve the desired effect. This is not conducive to the improvement of athletes’ skills. Therefore, this article aims to develop a big data motion target detection system based on deep convolutional neural network for sports difficult motion image recognition. More specifically, we use the high discriminative power of the convolutional neural network to extract images to perform computational preprocessing for the recognition of each human motion image in the video stream. Then, the skeleton recognition algorithm based on LSTM is used to detect the key points of the human body, which is of great significance for modeling different movements. Finally, we developed an object detection system to reconstruct each movement. By selecting five groups of highly difficult actions that are likely to cause sports injuries to conduct experimental research, the results prove the effectiveness of the target detection system we proposed.