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

Method for pedestrian weight recognition in video surveillance scene

A pedestrian re-identification and video surveillance technology, applied in the field of pedestrian re-identification, to achieve the effect of simple attribute classification and high accuracy

Inactive Publication Date: 2017-10-27
SOUTH CHINA UNIV OF TECH
View PDF5 Cites 34 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, due to changes in the viewing angle, illumination, distance, etc. of the surveillance video, the appearance of pedestrians often changes greatly, which brings great challenges to pedestrian re-identification.

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
  • Method for pedestrian weight recognition in video surveillance scene
  • Method for pedestrian weight recognition in video surveillance scene
  • Method for pedestrian weight recognition in video surveillance scene

Examples

Experimental program
Comparison scheme
Effect test

Embodiment

[0050] As shown in the attached picture, figure 1 The overall process of applying the method of this paper for pedestrian re-identification. Among them, the key steps involved in the present invention are training the CNN network and optimizing attribute weights, calculating attribute features, etc. The calculation and sequential output of the distance matrix are common steps for pedestrian re-identification applications.

[0051] figure 2 For this paper, the convolutional network used to extract deep features and obtain semantic attribute features is called FT-FNN (Fine-Tuning Feature Fusion Net). The upper part of the figure is the Alex network, and the lower part is a manually extracted feature ELF16, and the two are fused in the seventh fully connected layer. The number of nodes in the output layer of the network layer is consistent with the number of attributes to be identified.

[0052] as attached figure 1 As shown in, a method for pedestrian re-identification in a...

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 present invention discloses a method for pedestrian weight recognition in a video surveillance scene. The method is implemented through the following steps: S1, pre-training an FT-FNN network; S2, fine-tuning the FT-FNN network; S3, extracting a depth feature and an attribute feature of a training image; S4, optimizing the attribute weight; S5, extracting the attribute feature of a to-be-recognized image; S6, extracting the attribute feature of a pedestrian bank; S7, generating a distance matrix; and S8, outputting matching images in sequence. The depth feature after fine-tuning has strong distinguishing ability for the pedestrian weight recognition; compared with the texture, color and other low-level visual features, the middle-level semantic attributes are stable and not easy to change greatly with the change of the light and the posture; and manual integration of low-level vision features can improve the distinguishing ability of the depth feature to a certain extent, and can improve accuracy of some middle-level attributes closely related to the color and the texture, so that by combining the three advantages, the method disclosed by the present invention achieves better accuracy in the field of pedestrian weight recognition.

Description

technical field [0001] The invention relates to the technical field of computer vision, in particular to a method for pedestrian re-identification in a video surveillance scene. Background technique [0002] Person Re-identification (Person Re-identification) starts from multi-camera target tracking research, aiming to find specific target persons from other cameras. Pedestrian re-identification technology is mainly used in the field of security. With the popularization of video surveillance systems in recent years, video images have doubled. In the past, the reconnaissance method that relied entirely on manual screening of surveillance videos seemed inefficient. Therefore, the development of computers to retrieve matching targets from pedestrians The pedestrian re-identification technology of characters is particularly urgent. [0003] Due to the complex and uncontrollable monitoring environment, the changes in pedestrian postures and other conditions, the quality of pedes...

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): G06K9/00G06K9/46G06K9/62
CPCG06V20/53G06V10/443G06F18/214
Inventor 张见威邱隆庆林文钊
Owner SOUTH CHINA UNIV 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