In view of the current situation of the tracking system, which requires human control and is not intelligent enough, this paper designs an intelligent tracking system. The system uses the deep learning method to detect the objects, completes the analysis of the content of the image, and realizes intelligent tracking. Taking the curling movement as an example, the application of the intelligent tracking system is realized. In the experiment, the accuracy rate of the YOLO v2 model curling automatic shooting system detection data set was 94.3%, and the detection speed reached 30 f/s. The YOLO model can meet the real-time requirements of the curling and shooting system. In the actual process of shooting, the system was completed the tracking task.