Research and Implementation of Target Tracking Algorithm Based on Convolution Neural Network

摘要

Video target tracking is an important research topic in computer vision, and has been widely used in video surveillance, robot, human-computer interaction and so on. The emergence of large data age and the emergence of in-depth learning methods provide a new opportunity for the study of video target tracking. This paper first analyzes the research problems of video target tracking at present, analyzes the characteristics and trends of video target tracking in the new period, introduces the emerging recursive neural network frame structure, combined with Kalman filterAnd the experimental results show that the accuracy and robustness of the target tracking based on the convolution neural network algorithm are all good.