Pedestrian multi-target tracking method integrating target detection and association

Abstract

The invention discloses a pedestrian multi-target tracking method integrating target detection and association, and the method comprises the following steps: employing a training data set to train a tracking model network, and obtaining a tracking model; enabling a first frame of image in a to-be-tracked video stream to pass through the detection sub-model, and generating a boundary frame of a pedestrian target according to the heat map and the offset vector; extracting a feature vector of each pedestrian target through an appearance feature extraction sub-model, and distributing an ID and a track; enabling other frame images in the to-be-tracked video stream to pass through the tracking model in sequence; and according to the similarity of the feature vectors in the two adjacent frames ofimages in the to-be-tracked video stream, determining a corresponding track position of each pedestrian in the current frame of image, and connecting the track positions corresponding to the same IDin all frames of images in the to-be-tracked video stream to obtain a corresponding tracking result. The method overcomes the defects that a traditional tracking method is low in speed and frequent inidentity switching under the condition that an object is shielded.

Description

technical field [0001] The invention relates to the field of pedestrian tracking, in particular to a pedestrian multi-target tracking method integrating target detection and association. Background technique [0002] The main task of multi-target tracking is to find an image sequence, find the moving objects in the image sequence, and correspond the moving objects in different frames one by one, and then give the trajectory of different objects. These objects can be arbitrary, such as pedestrians, vehicles, athletes, various animals, etc., and the most studied is pedestrian tracking. This is because first of all, "pedestrians" are typical non-rigid objects, which are more difficult than rigid objects, and secondly, the detection and tracking of pedestrians has more commercial value in practical applications. According to incomplete statistics, at least 75% of the multi-target tracking research is on pedestrian tracking. [0003] Due to the development and application of co...

AI Technical Summary

Problems solved by technology

The common multi-target pedestrian tracking method in the existing technology is to divide the target detection and association into two steps, and use different data sets for training respectively. However, the problem caused by this method is that the running speed of the tracking model becomes slower, and it is difficult to achieve real-time tracking. Effect

Images