Video pedestrian detection method based on time-domain convolutional neural network

A convolutional neural network and pedestrian detection technology, applied in the field of intelligent video surveillance, can solve problems such as difficult to deal with pedestrian occlusion, difficult to detect pedestrians, etc., to achieve the effect of improving the detection rate and accuracy

Active Publication Date: 2017-08-04
BEIJING UNIV OF TECH
View PDF4 Cites 32 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0040] Aiming at the problem that the existing pedestrian detection method is difficult to deal with pedestrian occlusion, and it is difficult to detect pedestrians due to the interference of pedestrian posture, clothing, lighting and other factors, this invention proposes a pedestrian detection method based on deep convolutional neural network that considers time correlation. Detection Technology

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
  • Video pedestrian detection method based on time-domain convolutional neural network
  • Video pedestrian detection method based on time-domain convolutional neural network
  • Video pedestrian detection method based on time-domain convolutional neural network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0094] The present invention extracts 18 videos with complete pedestrian labels from the visual tracker benchmark database, decomposes and combines them into more than 10,000 samples, 60% of which are used for training, and 40% are used for testing.

[0095] The present invention adopts the method of "current frame + first four frames + last four frames" to obtain continuous frame images, with a total of 9 channels. In the screening stage of the predicted bounding box, the present invention compares the prediction result of the current frame with the detection results of the previous 4 frames. The overlap threshold of two bounding boxes is set to 0.7.

[0096] The main body of the convolutional network, referring to the Faster RCNN method [3], uses a network of 5 convolutional layers plus 2 fully connected layers. And in the training process, the network proposed by the present invention is initialized by utilizing the parameters of some layers trained by the Faster RCNN meth...

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 provides a pedestrian detection method based on a time-domain convolutional neural network. The pedestrian detection method of a single image is expanded for successive frames of images, the convolutional neural network is trained to learn the spatial correlation of the single video image and the time correlation between the successive frames, and the defect of detection of shielded pedestrians by the conventional method is overcome. Besides, pedestrian detection is conducted by employing the time-domain convolutional neural network so that the posture change for pedestrians is more robust, and the overall detection accuracy and the recall rate are both improved.

Description

technical field [0001] The invention belongs to the field of intelligent video monitoring and relates to a pedestrian detection method based on a time-domain convolutional neural network, which is particularly suitable for detecting pedestrians from video. Background technique [0002] With the development of science and technology and people's increasing emphasis on security, traditional video surveillance systems have become increasingly difficult to meet people's needs. The future video surveillance system will develop towards digitization, automation and intelligence. Not only to transmit surveillance images, but also to be able to detect and analyze pedestrians in the images. This project is dedicated to research on the key technology in the development of intelligent video surveillance system - pedestrian detection technology. [0003] The existing pedestrian detection for color images can basically be divided into two categories: methods based on background modeling...

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): G06K9/00G06K9/62G06N3/02
CPCG06N3/02G06V20/49G06F18/24G06F18/214
Inventor 胡永利冯乐乐孙艳丰尹宝才
Owner BEIJING UNIV OF 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