A pedestrian re-identification monitoring system based on a depth convolution neural network

A pedestrian re-identification and monitoring system technology, applied in the field of computer recognition, can solve problems such as difficult to capture samples, low resolution, difficult face recognition, etc., to achieve the effect of avoiding performance degradation, good training effect, and improving accuracy

Inactive Publication Date: 2018-12-14
中山大学新华学院
View PDF5 Cites 18 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] In the current technology, due to the generally low resolution of cameras, it is difficult to obtain recognizable features such as faces. In addition, due to the influence of illumination and viewing angles, the same pedestrian has great differences in different cameras. At the same time, high-dimensional The visual characteristics are usually difficult to capture the invariant factors of the sample, which leads to the shortcomings of the traditional pedestrian re-identification method, such as difficulty in identifying multiple pedestrians at the same time, time-consuming calculation, low recognition efficiency and correct rate, etc.

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
  • A pedestrian re-identification monitoring system based on a depth convolution neural network
  • A pedestrian re-identification monitoring system based on a depth convolution neural network
  • A pedestrian re-identification monitoring system based on a depth convolution neural network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0054] The present invention provides a pedestrian re-identification monitoring system based on deep convolutional neural network, such as figure 1 As shown, the system includes:

[0055] As the camera network of the video acquisition device, the camera network is formed by several high-speed network ball machines 11 applying WiFi and the switch 12; the video acquisition device is used to collect images of measured pedestrians;

[0056] As an identification and monitoring device, a main control computer 2 with an identification and monitoring function;

[0057] A storage server 3 storing a pedestrian database; the pedestrian database is used to store target pedestrian images;

[0058] The identification and monitoring device is used to obtain the image of the measured pedestrian from the video acquisition device; obtain the image of the target pedestrian from the database of pedestrians, and use the image of the target pedestrian to learn the characteristics of the target ped...

Embodiment 2

[0066] This embodiment is a preferred implementation mode based on the above-mentioned embodiment. The difference between this embodiment and the above-mentioned embodiment 1 is:

[0067] In this embodiment, the identification and monitoring device of the monitoring and identification system further includes an input device; the identification and monitoring device is also used to process the target pedestrian through natural language after obtaining the description of the target pedestrian from the input device. Describe and screen out the key information of the target pedestrian, and then use the key information of the target pedestrian to apply deep learning to obtain the characteristics of the target pedestrian. The key information of the target pedestrian refers to the color, gender, age group, etc. In this technical step, only the description of the relevant pedestrian needs to be input, and the relevant information will be automatically screened out after natural languag...

Embodiment 3

[0073] This embodiment is a preferred implementation mode based on the basic implementation mode of the above-mentioned embodiments. The difference between this embodiment and the above-mentioned embodiment 1 is that this embodiment is some examples and descriptions of the preferred implementation modes of the data set.

[0074] In some preferred embodiments, the datasets of the pedestrian database include public datasets with labels and real datasets without labels. When performing pedestrian re-identification training, preprocessed pedestrian images are required as training images and test images. Due to the high difficulty of sorting real data sets and the long period, in this technical solution, on the one hand, the existing public data sets are used for pedestrian re-identification training; Combined with training, the trained neural network can take various factors into consideration more comprehensively.

[0075] In some embodiments, the public data set includes one or...

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 monitoring system based on a depth convolution neural network. The system comprises a video acquisition device, an identification monitoring device and a pedestrian database. The video acquisition device is used for acquiring a pedestrian image to be measured; the pedestrian database is used for storing a target pedestrian image; the identification monitoring device is used for acquiring a measured pedestrian image from the video acquisition device, acquiring the target pedestrian image from the pedestrian database, learning the target pedestrian feature by using the target pedestrian image using a depth convolution neural network, and judging whether the measured pedestrian image is a target pedestrian according to the target pedestrian feature matching. The technical scheme can effectively reduce a plurality of adverse influence factors on identification robustness, thereby improving the matching accuracy of pedestrian target similarity.

Description

technical field [0001] The invention relates to the field of computer identification, in particular to a pedestrian re-identification monitoring system based on a deep convolutional neural network. Background technique [0002] The following statements merely provide background information related to the present disclosure and do not necessarily constitute prior art. [0003] Pedestrian re-identification refers to the process of similarity matching of pedestrian targets under the monitoring of multiple cameras without overlapping perspectives, that is, given a pedestrian target, find and lock the target in the videos shot by multiple cameras in different positions at different times . [0004] In the current technology, due to the generally low resolution of cameras, it is difficult to obtain recognizable features such as faces. In addition, due to the influence of illumination and viewing angle, the same pedestrian has great differences in different cameras. At the same ti...

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): G06K9/00G06K9/62G06N3/04G06N3/08
CPCG06N3/08G06V40/10G06V20/52G06N3/045G06F18/214G06F18/24
Inventor 瞿文政许志明王嘉茵肖泽彬廖嘉凯邱泽敏万智萍
Owner 中山大学新华学院
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