Dynamic Target Recognition and Tracking for Unmanned Aerial Vehicles Based on YOLOv5-DeepSORT

With the wide application of UAVs and the development of deep learning target detection algorithms represented by YOLOv5, multi-target dynamic tracking and detection systems supported by UAVs are increasingly available for design and application. By combining the YOLOv5 algorithm with the camera equipment and flight control of UAVs, designing a multi-target dynamic tracking and detection system can improve production efficiency and life. Given the disadvantages of YOLOv5, such as unstable counts in the processing of video streams, this paper combines the YOLOv5 algorithm with DeepSORT algorithm, which makes it possible to take into account the temporal influence of the image in the process of dynamic target recognition, and the error rate is reduced by 17.19% after the improvement, and the errors and omissions of detection have been significantly improved, and the effect of target tracking is obvious, which improves the accuracy of detection.