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

Training Sample Selection Method Based on Multi-Instance Learning Idea in Target Tracking

A multi-instance learning and target tracking technology, which is applied in image analysis, image enhancement, instruments, etc., can solve the problem that the sample label cannot be guaranteed

Active Publication Date: 2019-07-16
HARBIN INST OF TECH
View PDF4 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0007] The present invention aims to solve the defect of "weakly labeled" samples faced by traditional supervised learning algorithms in target tracking, resulting in the inability to guarantee accurate labeling of samples, thereby providing a training sample selection method based on the idea of ​​multi-instance learning in target tracking

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
  • Training Sample Selection Method Based on Multi-Instance Learning Idea in Target Tracking
  • Training Sample Selection Method Based on Multi-Instance Learning Idea in Target Tracking
  • Training Sample Selection Method Based on Multi-Instance Learning Idea in Target Tracking

Examples

Experimental program
Comparison scheme
Effect test

specific Embodiment approach 1

[0042] DETAILED DESCRIPTION 1. A method for selecting training samples based on the idea of ​​multi-instance learning in target tracking. The present invention proposes a method for selecting training samples based on the idea of ​​multiple instance learning (MIL) for target tracking. The main idea is to put all the positive samples into a positive sample package, and consider the samples that contribute less to the log-likelihood function of the package as the poor samples. In an iterative way, the worst samples are removed from the positive sample packet in each iteration until a sufficient number of samples remain in the positive sample packet.

[0043] The specific steps of the invention are:

[0044] Input: training set {(X 1 , Y 1 ),..., (X n , Y n )}, where X 1 ={x 11 ,..., x 1M } Is the first sample package, which is also a positive sample package, {X 2 ,..., X n } Is a negative sample package, y i ∈{0,1} is the label of the packet;

[0045] 1: Use the data set {(X 2 , Y 2 )...

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

A training sample selection method based on the idea of ​​multi-instance learning in target tracking relates to target tracking technology in the field of computer vision. The purpose of the present invention is to solve the defect of "weakly marked" samples faced by the traditional supervised learning algorithm in target tracking, resulting in the problem that the samples cannot be accurately marked. This method: put all positive samples into a positive sample bag, and regard samples that contribute less to the log-likelihood function about the bag as poor samples. In an iterative manner, each iteration removes the worst sample from the positive sample bag until a sufficient number of samples remain in the positive sample bag. The invention is superior to other evaluation algorithms in terms of tracking accuracy, has a higher frame rate and meets real-time requirements.

Description

Technical field [0001] The invention relates to target tracking technology in the field of computer vision. Background technique [0002] Although tracking algorithms for specific objects such as human faces and pedestrians have achieved a series of successes, due to the lack of sufficient prior information for an arbitrary object, the appearance of the object will occur due to deformation, rotation or changes in illumination during the tracking process. Due to drastic changes and other reasons, the target tracking algorithm for general objects still faces many challenges. Because the tracking algorithms for general targets are not reliable enough, the development of many application scenarios based on general tracking is limited. For example, the target tracking function in the camera module of a smart phone often requires the target to have obvious characteristics, or it can only achieve a short-term tracking of the target. [0003] Although more and more researchers believe th...

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): G06T7/246G06K9/62
CPCG06T2207/30232G06T2207/10016G06T2207/20081G06F18/24155G06F18/214
Inventor 贾敏高政王雪高天娇尹志胜郭庆顾学迈
Owner HARBIN INST OF TECH
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