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

A face detection method based on convolution neural network cascade for large-scale scene monitoring images

A convolutional neural network and face detection technology, which is applied in the field of face detection in large scene surveillance images, can solve the problems of poor face detection effect and achieve high detection accuracy and high detection accuracy

Active Publication Date: 2018-12-14
SHANGHAI UNIV
View PDF6 Cites 4 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] In order to overcome the deficiencies of the above-mentioned prior art, the present invention provides a large-scene monitoring image face detection method based on convolutional neural network cascading, to solve the problems of existing face detection algorithms for faces in large-scene monitoring images. The problem of poor detection effect

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 face detection method based on convolution neural network cascade for large-scale scene monitoring images
  • A face detection method based on convolution neural network cascade for large-scale scene monitoring images
  • A face detection method based on convolution neural network cascade for large-scale scene monitoring images

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0032] Implementation example 1: refer to the attached Figure 1~5 :

[0033] (1-1) The large-scene surveillance image face detection method based on the convolutional neural network cascade collects and organizes the large-scene surveillance image data and makes detailed annotations;

[0034] (1-2) Integrate the face area and its semantic information, that is, the cascaded convolutional neural network structure design of the face and its surrounding shoulder area;

[0035] Cascaded Convolutional Neural Network Model Training Integrating Face Regions and Their Semantic Information.

Embodiment 2

[0036] Implementation example 2: This implementation example is basically the same as implementation example 1, and the special features are as follows:

[0037] (2-1) The large scene monitoring image face detection method based on convolutional neural network cascading collects the monitoring image data of the large scene area of ​​a square, or a station, or a stadium;

[0038] (2-2) Use the labelImg image annotation software to annotate the image data collected in step (2-1), including the "rectangular area position of the face" and "semantic information of the face - the rectangular area position of the face and its surrounding shoulder area" ", generate an XML file after marking;

[0039] (2-3) For the XML file generated after labeling, use Python to write a program to convert the label position coordinates saved in the XML file into the corresponding text file, paying special attention to comparing the position coordinates of the face area with the corresponding face sema...

Embodiment 3

[0044] Implementation example three: the face detection method of large scene monitoring images based on convolutional neural network cascading is characterized in that it includes the following steps:

[0045] (3-1) In the present invention, the training data of the algorithm model is obtained by collecting large-scale scene monitoring image data, and a total of 500 large-scale scene monitoring images are collected, and the involved scenes include squares, stations, and stadiums;

[0046] (3-2) A total of about 15,000 faces are marked in the collected images, and the semantic information areas corresponding to the faces are marked. For the geometric relationship between the face area and its corresponding semantic information area when the monitoring image data is annotated, refer to the attached figure 1 ;

[0047] (3-3) Integrate the face area and its semantic information, that is, the cascaded convolutional neural network structure design of the face and its surrounding s...

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 face detection method based on convolution neural network cascade for large-scale scene monitoring images. The specific operation steps of the method are as follows: (1) collecting and sorting large-scale scene monitoring image data and making detailed labeling; (2) designing a cascading convolution neural network structure by fusing the face region and its semantic information, that is, the face and its surrounding shoulder region; (3) training the cascading convolution neural network model which fuses face region and its semantic information. The invention effectively solves the problem of poor face detection effect caused by small face scale, large scale change and blurred face details in the large-scale scene monitoring image, and the detection performance isgreatly improved compared with the existing face detection algorithm.

Description

technical field [0001] The invention relates to the fields of computer vision and artificial intelligence, in particular to a face detection method for large scene monitoring images based on convolutional neural network cascading. Background technique [0002] "Face detection" is one of the classic computer vision research topics. With the advancement of the times, "face detection" has become more and more widely used in our lives. For these daily applications, there are already many mature algorithms, such as the open source Opencv face detection algorithm, Dlib These traditional face detection algorithms such as face detection algorithms. [0003] Face detection algorithms based on deep learning, such as CascadeCNN and MTCNN, have emerged in recent years. Thanks to the powerful learning ability of the deep learning algorithm, the performance of the face detection algorithm is greatly improved. [0004] In recent years, the international security situation is not optimis...

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
CPCG06N3/08G06V40/172G06V40/168G06N3/045G06F18/2148G06F18/24
Inventor 卜伟周传宏
Owner SHANGHAI UNIV
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