Target tracking method and device

Abstract

The present invention provides a target tracking method and device, which are applied to a server storing feature models corresponding to each target object. The method includes: performing target detection on the current video frame image, and extracting corresponding CNN features according to the detected position information of each object to be detected; The position information and feature model of each target object in the image are calculated to obtain the corresponding similarity matrix; based on the similarity matrix, the data association between each object to be detected and each target object is carried out to obtain the optimal matching result; if there is When the object to be detected is successfully matched with the corresponding target object, the corresponding feature model is updated according to the CNN features of the object to be detected, and a corresponding tracking result is obtained based on the object to be detected. The method has strong anti-interference ability, high tracking success rate, and can continuously track the target object.

Description

technical field [0001] The present invention relates to the technical field of multi-target tracking of video images, in particular to a target tracking method and device. Background technique [0002] With the continuous development of surveillance technology, the application of multi-target tracking technology for tracking multiple target objects in surveillance video is becoming more and more extensive. In the process of tracking the target object, the existing multi-target tracking scheme combines the CNN (Convolutional Neural Network, Convolutional Neural Network) feature of the target object in the current video image with the CNN feature when the target object was successfully tracked recently. The comparison method is used to track the target object, but this multi-target tracking solution has weak anti-interference ability and low target tracking success rate, usually because the CNN feature of the target object recently successfully tracked carries some occlusions ...

AI Technical Summary

Problems solved by technology

In the process of tracking the target object, the existing multi-target tracking scheme combines the CNN (Convolutional Neural Network, Convolutional Neural Network) feature of the target object in the current video image with the CNN feature when the target object was successfully tracked recently. The comparison method is used to track the target object, but this multi-target tracking solution has weak anti-interference ability and low target tracking success rate, usually because the CNN feature of the target object recently successfully tracked carries some occlusions , so that the CNN features of the target object in the current video image cannot be correctly matched with the CNN features of the target object when it was successfully tracked recently, resulting in tracking failure

Images