Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Target tracking method based on enriched target form change updating template

A morphological change, target template technology, applied in the field of target tracking, can solve problems such as difficult to deal with target occlusion and rapid deformation, achieve good objective evaluation results and subjective results, improve the accuracy of matching and calculation, and enhance adaptability. Effect

Active Publication Date: 2020-08-25
TIANJIN UNIV
View PDF3 Cites 7 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] Due to the shape change of the target during the tracking process, most of the tracking algorithms under the framework of the deep twin network are offline tracking, which is difficult to deal with the problems of the target being occluded and rapidly deformed during the tracking process.

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
  • Target tracking method based on enriched target form change updating template
  • Target tracking method based on enriched target form change updating template
  • Target tracking method based on enriched target form change updating template

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0031] The embodiment of the present invention proposes a target tracking method based on enrichment target morphological change update template, see figure 1 , the method includes the following steps:

[0032] 101: Build the basic network framework of the target tracking algorithm: that is, first build the SiameseRPN++ based [8] The basic tracking framework, and then introduce the FlowNet-based [9] The optical flow extraction and mapping modules form a complete basic network framework.

[0033] Among them, FlowNet [9] Contains a stack of three-layer optical flow extraction networks, the present invention uses one of the modules, FlowNetC [10] As the basic network of optical flow extraction, the optical flow information between the first frame target template and the latest frame target template is extracted during the tracking process to describe the shape changes of the target during the tracking process.

[0034] 102: In the tracking process, the present invention adds ...

Embodiment 2

[0040] The scheme in embodiment 1 is further introduced below, see the following description for details:

[0041]201: In the traditional target tracking network based on the deep twin network framework, the input of the target template branch is usually relatively single, that is, it is always the target template of the first frame after initialization. In the calculation of subsequent frames of the video, this template and The search area cut out in the next frame is used for feature matching and calculation. During the tracking process, if the target changes rapidly, is occluded, or deforms, the information in the target template in the first frame often has poor timeliness and is quite different from the target in the current frame, which will lead to deviation of the tracking results. Even tracking fails. Therefore, the present invention adds a target template online update mechanism in the basic network framework, adopts a method of enriching target shape changes, and f...

Embodiment 3

[0061] Below in conjunction with concrete experimental data, the scheme in embodiment 1 and 2 is carried out effect assessment, see the following description for details:

[0062] 301: Data composition

[0063] The test set consists of all video sequences (60 in total) in the VOT2016 database.

[0064] 302: Evaluation Criteria

[0065] The present invention mainly uses three evaluation indexes to evaluate the performance of the target tracking algorithm:

[0066] Accuracy (accuracy rate) is an indicator to measure the tracking accuracy of the tracking algorithm. By calculating the IoU (overlap rate, Intersection over Union) between the predicted target regression frame and the real target regression frame for each frame in the video sequence,

[0067]

[0068] in, represents the ground truth target regression box, Represents the predicted target regression box. In order to ensure the accuracy, the accuracy of each frame will be measured repeatedly, and all the resul...

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 a target tracking method based on an enriched target form change updating template. The target tracking method comprises the following steps: constructing a basic network framework for target tracking; adopting a mapping function in the basic network framework and a mapping function based on bilinear interpolation to input the first frame of target template and optical flowinformation between the first frame of target template and the nearest frame of target template, and obtaining a nearest frame of target mapping template; calculating a residual error between the latest frame of target mapping template and the first frame of target template, obtaining form change information of the target from the first frame to the latest frame, linearly weighting the form change information to obtain a residual error graph of the latest form change of the target, and adding the residual error graph to the first frame of target template according to pixels to obtain a current frame of target template with enriched target deformation information; and inputting the current frame target template into the feature extraction network, calculating the position offset and the size of the next frame target, and completing the tracking of the next frame target. According to the method, template updating is carried out frame by frame, and challenging problems of rapid change, shielding, target deformation and the like in the tracking process are effectively solved.

Description

technical field [0001] The invention relates to the field of target tracking, in particular to a target tracking method based on enriching target shape changes and updating templates under the framework of a deep twin convolutional neural network. Background technique [0002] With the development and penetration of the field of computer vision, more and more artificial intelligence and automation technologies have entered people's lives. As a very important research direction in computer vision, object tracking has attracted the attention of many researchers at home and abroad for a long time. At present, the target tracking algorithm has been widely used in many fields such as automatic driving, pedestrian detection, human-computer interaction and smart city, and has broad development prospects. [0003] Although the target tracking technology has developed rapidly, due to the variability and complexity of the application scenarios, there are still many challenging proble...

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): G06T7/246G06T7/223G06N3/04G06N3/08
CPCG06T7/246G06T7/223G06N3/08G06N3/045
Inventor 张静郝志晖刘婧苏育挺
Owner TIANJIN UNIV
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

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

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com