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

Multi-target model visual tracking method based on cost-sensitive three-way decision-making

A multi-objective model and cost-sensitive technology, applied in the field of surveillance video intelligent analysis, can solve problems such as lack of multi-features, multi-decision types, poor accuracy and robustness, etc., achieve accurate and robust tracking effects, and improve The effect of accuracy

Active Publication Date: 2022-05-13
TONGJI UNIV
View PDF10 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

Although the addition of spectral information helps to improve the tracking effect, the traditional method does not have the advantages of multi-features and multi-decision types, and its accuracy and robustness are not good.

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
  • Multi-target model visual tracking method based on cost-sensitive three-way decision-making
  • Multi-target model visual tracking method based on cost-sensitive three-way decision-making
  • Multi-target model visual tracking method based on cost-sensitive three-way decision-making

Examples

Experimental program
Comparison scheme
Effect test

Embodiment

[0048] In order to make the object of the present invention, the technical solution and advantages more clearly understood, the following combined embodiments, as specific as Figure 1 Algorithm flowchart shown, the present invention is further elaborated in detail. It should be understood that the specific embodiments described herein are only used to explain the present invention, but do not limit the present invention.

[0049] Step 1: Establish multiple target models with different update strategies, which are described as follows: The target model should have the following characteristics: keep the characteristics of the tracked target unchanged, adapt to the appearance changes of the target in time, and be able to predict the future appearance changes of the target. Thus, the present invention first saves the target appearance of the initial frame, as a target feature that remains unchanged; secondly, the target appearance will be updated frame by frame, as a target feature ...

specific Embodiment approach

[0067] Figure 1 is an algorithm implementation flowchart of the present invention, the specific embodiments are as follows:

[0068] 1. Establish multiple target models with different update strategies and form a collection

[0069] 2. Use similarity learning model tracking to obtain multi-model candidate boxes ij for subsequent calculations of Articles 3 and 6;

[0070] 3. Calculate the number N of all candidate boxes that overlap in two pairs t ；

[0071] 4. Set a cost dictionary for each decision type, according to N t Calculate the cost of decision-making;

[0072] 5. Based on the cost-sensitive theory of three-branch decision-making, calculate the upper and lower boundaries of the three decision-making α and β according to the decision-making cost;

[0073] 6. According to the upper and lower boundary α and β, the candidate box obtained in step 2 is classified by using the convolutional neural network discriminator;

[0074] 7. The candidate box with the highest confidence ...

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 relates to a multi-target model visual tracking method based on cost-sensitive three-way decision-making, comprising the following steps: 1) establishing a multi-target model Z in the i-th frame i ; 2) Using the multi-objective model Z i A multi-target set x is found in frame (i+1) i+1 ; 3) According to the multi-target set x i+1 The position overlap of the samples in the middle is calculated respectively to obtain the cost value λ of the six decision-making outcomes of positive-acceptance, positive-rejection, negative-acceptance, negative-rejection, intermediate-acceptance, and intermediate-rejection; 4) Based on the cost-sensitive The three-way decision-making method obtains the decision boundary and sets the multi-objective x i+1 Divided into positive samples, negative samples and intermediate samples 5) Select the result with the highest confidence s (i+1)1 , as a temporary tracking result, and the recorded tracking result is reserved; 6) In the (i+2)th frame, if s exists in (i+1)j can provide more reliable tracking results, replace it with s (i+1)1 Provided tracking results. Compared with the prior art, the present invention has the advantages of rapidity, accuracy, good robustness and the like.

Description

Technical field [0001] The present invention relates to the field of intelligent analysis of surveillance video, in particular to a multi-objective model visual tracking method based on a cost-sensitive three-branch decision. Background [0002] Visual tracking refers to the process of determining the target for any category in a short video (generally less than 1000 frames, and there is no lens switching), after the first frame is given its position and size, and then monitoring its position and size online in subsequent frames. Compared with the target detection algorithm, this algorithm generally has the characteristics of fast speed and high accuracy, and is mainly used in the following scenarios: (1) scenes that need to identify the trajectory of the target motion; (2) candidate box labeling for auxiliary target detection; (3) cooperate with the target recognition algorithm to extract candidate targets, save computing power, and achieve efficient and accurate identification...

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 Patents(China)
IPC IPC(8): G06V20/40G06V10/62G06V10/764G06V10/82G06V10/74
CPCG06V20/41G06V2201/07G06F18/241
Inventor 赵才荣孙添力
Owner TONGJI 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