Target and Recognition Tracking Method Based on Yolov3 Network and Mean Shift

Abstract

The present invention is a target and identification tracking method based on the YOLOv3 network and mean shift, comprising the following steps: S1: making a target data set, and inputting the target data set into the YOLOv3 algorithm training network; S2: inputting the video images into the YOLOv3 network frame by frame, After the detection is completed, the YOLOv3 network outputs the detection target prediction label vector; S3: Input the multiple target bounding box size information in the detection target prediction label vector into the mean-shift target tracking algorithm, and use the bounding box size information of each target as the mean-shift target tracking. The algorithm corresponds to the size of the search detection frame of the target; S4: Track the target; S5: Determine whether the video ends. The method solves the problem of the mismatch of the scale of the detection frame searched by the traditional mean-shift target tracking algorithm, improves the limitation of the traditional mean-shift target tracking algorithm, and improves the performance of the target recognition and tracking algorithm.

Description

technical field [0001] The invention belongs to the technical field of computer target detection and target tracking algorithms, in particular to the research on a target recognition and tracking method based on YOLOv3 network and mean shift. Background technique [0002] The expansion of urban traffic scale and the improvement of supporting facilities are one of the important characteristics of my country's rapid development. At the same time, the continuous development of urban traffic has also improved people's work efficiency and made people's life and travel easier and more convenient. The number of motor vehicles has exploded. The explosive increase in the number of motor vehicles has brought serious congestion problems. People's travel convenience has been greatly challenged. Road congestion has also increased energy consumption and exhaust pollution. Therefore, people's demand for improving the problem of traffic congestion is also increasing. [0003] Combining compu...

AI Technical Summary

Problems solved by technology

[0007] In order to solve the above problems, the present invention provides a target and recognition tracking method based on YOLOv3 network and mean shift, which solves the problem of scale mismatch in the search detection frame of the traditional mean shift target tracking algorithm, so that it can run efficiently in ITS Computer Target Recognition and Tracking Algorithm

Images