Unsupervised pedestrian re-identification method based on multi-feature clustering

A pedestrian re-identification and multi-feature technology, which is applied in the field of unsupervised pedestrian re-identification based on multi-feature clustering, can solve the problems of poor robustness of pseudo-label prediction algorithms, increase intra-class similarity, reduce intra-class variance, The effect of improving robustness

Pending Publication Date: 2022-07-29
苗满田
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0006] The purpose of the present invention is to overcome the deficiencies of existing methods, to design an unsupervised pedestrian re-identification method, especially to design an unsupervised pedestrian re-identification method based on multiple feature clustering, to solve the problem of the existing pseudo-label prediction algorithm. The problem of poor stickiness

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
  • Unsupervised pedestrian re-identification method based on multi-feature clustering
  • Unsupervised pedestrian re-identification method based on multi-feature clustering
  • Unsupervised pedestrian re-identification method based on multi-feature clustering

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0039] The technical solutions in the embodiments of the present invention will be described below with reference to the accompanying drawings in the embodiments of the present invention.

[0040] figure 1 is the specific flow chart of the implementation of the present invention, figure 2 is the overall frame diagram of the present invention, such as figure 1 and figure 2 As shown, the method includes:

[0041] Step 1: Divide the training set based on the camera label, train the feature extraction model inside each camera based on the identity loss, and obtain the feature extraction model for different camera styles;

[0042] Step 2: Feed each sample into each feature extraction model, and for each sample, splicing all its features, use the spliced ​​features to cluster and generate pseudo-labels;

[0043] Step 3: Based on pseudo-labels, the shared feature extraction model is optimized in two stages to increase intra-class similarity and reduce intra-class variance.

[...

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 an unsupervised pedestrian re-identification method based on multi-feature clustering. Comprising the following steps: 1, dividing a training set based on camera tags, and training a feature extraction model in each camera based on identity loss to obtain feature extraction models for different camera styles; 2, feeding each sample into each feature extraction model, splicing all features of each sample, clustering by using the spliced features, and generating a pseudo tag; and 3, based on the pseudo tag, optimizing the shared feature extraction model in two stages so as to increase the intra-class similarity and reduce the intra-class variance. The unsupervised pedestrian re-identification method based on multi-feature clustering designed by the invention aims to reduce the influence of camera difference on a clustering result, so that the identification precision of the model is improved.

Description

Technical field: [0001] The invention relates to the field of pedestrian re-identification, in particular to an unsupervised pedestrian re-identification method based on multiple feature clustering. Background technique: [0002] The purpose of pedestrian re-identification is to retrieve images of pedestrians with specified identities from a library of images captured by multiple cameras. With the rise of deep convolutional neural networks, supervised learning-based person re-identification methods have achieved satisfactory performance in recent years. To alleviate the expensive annotation cost, researchers in this field have gradually turned their attention to unsupervised person re-identification. [0003] Existing related work on unsupervised person re-identification can be roughly classified into three categories: 1) At the pixel level, Generative Adversarial Network (GAN) is applied for image style transfer while maintaining identity annotations on the source domain; ...

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): G06V10/774G06V10/762G06V20/52G06V40/10
CPCG06F18/23G06F18/214
Inventor 苗满田
Owner 苗满田
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