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.

对篮球高难度动作识别技术的分析，主要是对视频中篮球运动员的身体行为进行识别和分析，以完成技术动作。视频识别的目的是为提高篮球训练水平提供重要保证。当前的目标识别技术已经取得了一些成果。结果表明，目标检测技术在篮球运动场景中的应用具有重要意义，可以提高运动训练的效果。但是，传统的运动目标识别受技术和伤害的限制，对难运动技能的分析受场景，动态背景和技术的限制，无法达到预期的效果。这不利于运动员技能的提高。因此，本文旨在开发一种基于深度卷积神经网络的大数据运动目标检测系统，用于运动困难的运动图像识别。更具体地说，我们使用卷积神经网络的高判别力来提取图像，以执行计算预处理，以识别视频流中的每个人体运动图像。然后，基于LSTM的骨骼识别算法被用于检测人体的关键点，这对于建模不同的动作具有重要意义。最后，我们开发了一种物体检测系统来重构每个运动。通过选择五组可能造成运动伤害的高难度动作进行实验研究，结果证明了我们提出的目标检测系统的有效性。