A multi-target tracking method and system based on graph matching

Abstract

The invention discloses an online multi-target tracking algorithm based on graph matching, which transforms the data association problem of detection responses between two consecutive frames into a graph matching problem. Firstly, two deep convolutional neural networks are designed to solve the intimacy between the vertices of the two graphs and the intimacy between the edges of the two graphs respectively, and then directly use the intimacy between the vertices and the intimacy between the edges to fill the two The intimacy matrix between two graphs, and finally the intimacy matrix is ​​processed to obtain the final matching matrix (that is, the correlation matrix between detections). Therefore, the present invention can effectively reflect the correlation of real data in the process of multi-target tracking, and the accuracy of tracking results is high.

Description

technical field [0001] The invention belongs to the technical field of pattern recognition, and more particularly relates to a multi-target tracking method and system based on graph matching. Background technique [0002] Multiple Object Tracking (MOT) plays an important role in the field of computer vision, and its main task is to analyze videos to identify and track objects belonging to one or more categories without any prior knowledge about appearance and number of objects. It plays an important role in the fields of motion analysis, human-computer interaction, video surveillance (such as abnormal behavior recognition) and automatic driving. [0003] The existing multi-target tracking algorithm mainly includes three methods. The first one is to use the first detection and then tracking strategy, that is, the detection-based tracking paradigm first uses the detector to locate the position of the target of interest in each frame, and then passes the data Association assig...

AI Technical Summary

Problems solved by technology

[0008] In view of the above defects or improvement needs of the prior art, the present invention provides a multi-target tracking method and system based on graph matching, the purpose of which is to solve the detection between consecutive frames by combining deep learning with the traditional graph matching framework The problem of data association between them, so as to complete the online multi-target tracking, and further solve the technical problem that the topological relationship between vertices in the existing multi-target tracking method is not fully utilized, and the multi-target tracking method based on the graph model is due to The technical problem of low target tracking accuracy due to fixed structure, and the technical problem of manual design parameters due to the high complexity of the machine learning model in the multi-target tracking method using the machine learning model

Images