Image pedestrian re-identification method and system based on multi-attention joint learning

A pedestrian re-identification and attention technology, applied in the field of pedestrian re-identification research, can solve the problem of low accuracy of pedestrian re-identification

Active Publication Date: 2020-08-14
HUAZHONG UNIV OF SCI & TECH
View PDF4 Cites 21 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] Aiming at the defect of low pedestrian re-identification accuracy caused by the use of single attention in the prior art and the need for improvement, the present invention provides a method and system for image pedestrian re-identification based o

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
  • Image pedestrian re-identification method and system based on multi-attention joint learning
  • Image pedestrian re-identification method and system based on multi-attention joint learning
  • Image pedestrian re-identification method and system based on multi-attention joint learning

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0050] The present invention simultaneously introduces the Soft attention module and the high-order attention module into the ResNet50 network to form a multi-attention joint learning network, initializes the multi-attention joint learning network with the pre-trained ResNet50 network parameters, and then in the Market-1501 data set After training, the network can extract effective pedestrian representation features for pedestrian re-identification.

[0051] The invention discloses an image pedestrian re-identification method based on multi-attention joint learning, which includes the following steps:

[0052] Step 1. Pre-train the ResNet50 network so that the ResNet50 network parameters have initial values.

[0053] Get the ImageNet dataset, the URL of the dataset is https: / / www.image-net.org / , use the amsgrad algorithm to update the network parameters, the formula of the amsgrad algorithm is:

[0054] m t = β 1 m t-1 +(1-β 1 )g t

[0055]

[0056]

[0057]

...

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 image pedestrian re-identification method and system based on multi-attention joint learning, and belongs to the technical field of image processing. Soft attention and high-order attention are introduced into a ResNet50 feature extraction network; by utilizing the complementary effect of two different types of attention on feature extraction, the learning ability of thefeature extraction network on pedestrian features is improved, so that the feature extraction network pays attention to more discriminative features in pedestrian images. In order to obtain more accurate attention features, a multistage attention loss function is provided, the loss function is utilized to guide training and learning of the feature extraction network, and the learning ability of the feature extraction network for pedestrian features is further improved. When pedestrian global features are learned, intermediate features in the feature extraction network are fused, learning of pedestrian local features is enhanced, the ability of the network to learn subtle differences between pedestrian features is improved, and the performance of the network in image pedestrian re-identification is improved.

Description

technical field [0001] The invention belongs to the research field of pedestrian re-identification in image processing and machine vision, and in particular relates to an image pedestrian re-identification method and system based on multi-attention joint learning. Background technique [0002] Pedestrian re-identification is a fundamental task in automatic video surveillance, and it is also a research hotspot in recent years. The purpose of person re-identification is to establish corresponding links between observations of the same person under different cameras. Typically, this is done by taking an image (or set of images) of a person seen in one camera view and forming a descriptive model for comparison with images of pedestrians observed in another camera view or point in time . Its purpose is to determine the past (or present) position of a person in a set of cameras by finding the correct matching image. [0003] Person re-identification is a very difficult research...

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
IPC IPC(8): G06K9/00G06K9/62G06N3/04
CPCG06V40/103G06N3/048G06N3/045G06F18/22G06F18/214
Inventor 韩守东罗善益张宏亮刘东海生
Owner HUAZHONG UNIV OF SCI & TECH
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