Pedestrian re-identification method using attitude information to design multi-loss function

A pedestrian re-identification, loss function technology, applied in the field of computer vision

Active Publication Date: 2018-03-23
BEIHANG UNIV
View PDF5 Cites 107 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, only using the verification model can only

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
  • Pedestrian re-identification method using attitude information to design multi-loss function
  • Pedestrian re-identification method using attitude information to design multi-loss function
  • Pedestrian re-identification method using attitude information to design multi-loss function

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0088] The specific steps of the present invention will be described in detail below in conjunction with the accompanying drawings and embodiments.

[0089] The present invention proposes a pedestrian re-identification method using attitude information to design multiple loss functions, first combining figure 1 The overall schematic diagram introduces the pedestrian re-identification process of the present invention in detail. The present invention includes offline and online stages, wherein the offline stage includes preprocessing, rough feature extraction, fine feature extraction, feature fusion, quintuple similarity measurement, multi-class loss function calculation, and network parameter learning; the online stage is It includes four parts: feature extraction, similarity measurement, image library update and result visualization.

[0090] Stage (1) Offline stage: training and learning the network model for extracting features.

[0091] A. The steps of data preprocessing ...

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 pedestrian re-identification method using attitude information to design a multi-loss function. The method can effectively solve the difficulties caused by frequent occlusionof pedestrians, large differences of video illumination and non-rigid pedestrian attitudes in surveillance video, and has a wide range of applications in security monitoring and other fields. The method is mainly divided into two phases, which are an offline phase and an online phase. The offline phase is responsible for training and learning a high-accuracy deep learning network model, and includes preprocessing, extraction of joint point information, extraction of local features and feature fusion with global features extracted by a backbone network framework, and finally training of the fused features using a quintuple loss function. In the online phase, the trained deep learning network model is used for feature extraction, so that pedestrian re-identification between a target to be analyzed and a stored target picture library is achieved by similarity calculation.

Description

technical field [0001] The invention belongs to the field of computer vision technology, and more specifically, relates to a pedestrian re-identification method using posture information to design multiple loss functions. Pedestrian re-identification method. Background technique [0002] Pedestrian re-identification technology refers to retrieving a given target in multiple cameras, and correlating and matching the retrieval results. This technology provides basic support for applications in the field of video surveillance, such as pedestrian retrieval, cross-camera tracking, and human-computer interaction. For the person search task of massive video data, pedestrian re-identification can greatly liberate manpower. However, due to different camera viewing angles, complex lighting conditions, frequent occlusions, and variable poses of non-rigid pedestrians, the problem of pedestrian re-identification is very challenging. To overcome these difficulties, researchers have pro...

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): G06K9/00G06K9/62
CPCG06V40/103G06F18/2431G06F18/22G06F18/253G06F18/214
Inventor 周忠吴威姜那刘俊琦孙晨新
Owner BEIHANG UNIV
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