Online multi-target tracking method based on track metric learning

A multi-target tracking and metric learning technology, applied in the field of online multi-target tracking, which can solve problems such as poor practicability

Active Publication Date: 2019-06-25
NORTHWESTERN POLYTECHNICAL UNIV
View PDF23 Cites 8 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] In order to overcome the shortcomings of poor practicability of existing online multi-target tracking methods, the present invention provides an online multi-target tracking method based on trajectory metric learning

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Online multi-target tracking method based on track metric learning
  • Online multi-target tracking method based on track metric learning
  • Online multi-target tracking method based on track metric learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0051] refer to Figure 1-3 . The specific steps of the online multi-target tracking method based on trajectory metric learning in the present invention are as follows:

[0052] Step 1. Consider the length of the trajectory, the degree of occlusion of the target, the closeness of the trajectory and the detection response, the smoothness of the trajectory and other factors, and design the trajectory confidence function:

[0053]

[0054] in Indicates the trajectory length of the i-th target up to time t, Indicates the closeness between the i-th target and its corresponding detection response at time k, Indicates the unobstructed degree of the i-th target at time k, T k is all target trajectories at time k, smo(T t i ) represents the smoothness of the trajectory of the i-th target up to time t.

[0055] Step 2. For the trajectory T of the i-th target at time t t i , use the Kalman filter to estimate the position of the target at time k Considering the visual and...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention discloses an online multi-target tracking method based on track metric learning. The method is used for solving the technical problem that an existing online multi-target tracking methodis poor in practicability. According to the technical scheme, the method comprises: firstly, an existing target detection algorithm is used for generating a detection response; and then, dividing theexisting track set into a high-confidence set and a low-confidence set, processing a data connection problem of the high-confidence set and the next moment detection response by using static featuresand a traditional measurement method, and enhancing a data connection capability for the low-confidence set by using a similarity measurement matrix to obtain a final result. Existing track information is taken as a training sample set, a similarity measurement matrix between tracks and detection responses is learned online, and the resolution capability of track discrimination is enhanced. The technical problems that accurate data connection is difficult to perform and the tracking effect is seriously limited by the detection effect in the background technology are solved, the multi-target tracking effect is improved, and the practicability is good.

Description

technical field [0001] The invention relates to an online multi-target tracking method, in particular to an online multi-target tracking method based on trajectory metric learning. Background technique [0002] Visual multi-target tracking can match the detection responses at different times one by one under the given video sequence and target detection results, and finally obtain the dynamic changes of the target in the time and space domains. As an important part of visual perception and understanding, multi-target tracking can effectively extract target motion information and timing change information, and can assist smart devices to accurately and robustly understand the surrounding environment in time and space domains. For example: applying multi-target tracking to intelligent monitoring systems will help to detect abnormal targets; applying to smart cars will help avoid traffic accidents such as rear-end collisions and collisions; applying to smart robots will help T...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
Patent Type & Authority Applications(China)
IPC IPC(8): G06T7/246G06K9/46
Inventor 王琦李学龙张星宇
Owner NORTHWESTERN POLYTECHNICAL UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap