Deep network-based multi-strategy global crowd analysis method

A technology of deep network and analysis method, applied in the field of machine vision artificial intelligence, which can solve problems such as background interference and pedestrian occlusion

Inactive Publication Date: 2018-10-30
苏州平江历史街区保护整治有限责任公司
View PDF5 Cites 35 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0010] The technical problem to be solved by the present invention is to provide a deep network-based multi-strategy global crowd analysis method to overcome the background interference of complex scenes and pedestrian occlusion problems, and then realize the crowd density analysis in the scene. an accurate estimate of

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
  • Deep network-based multi-strategy global crowd analysis method
  • Deep network-based multi-strategy global crowd analysis method

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0039] The present invention will be described in more detail below in conjunction with the accompanying drawings and embodiments.

[0040] S1. Data preparation

[0041] For the crowd pictures in the same scene, about 1500 frames were intercepted from the surveillance video, including pictures with different lights and different numbers of people from 6:30-17:00. Use the tool to mark the pedestrian's head in each frame to generate the real point set data of each crowd position. To mark the crowd individuals in each frame, the acquisition method is to mark the fixed points on the head of the person, and mark the sparsely identifiable pedestrian head or the image of the complete torso with a mark frame. Gaussian convolution is performed on the marker map positions to generate an approximate density map. x i Indicates the head marker position x i , δ(x-x i) represents the impact function of the head position, N represents the total number of people, and G is the Gaussian ke...

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 deep network-based multi-strategy global crowd analysis method. The method comprises the following steps of: firstly, monitor area modeling: drawing a global map schematic diagram, establishing a layer for a direction and a range corresponding to a global map in a camera monitor area, and waiting for import of crowd density data; secondly, for a monitor scene of each camera, obtaining a space visual angle mapping graph of a displayed monitor image through perspective transformation, namely, overlook visual angle mapping from side-looking visual angle of the camera to the ground; obtaining image features through a VGG16 transfer learning method, mapping pre-blocks input into the images to a feature layer through strides, carrying out SWITCH judgement on the image features of each block, and selecting to carry out density estimation or pedestrian detection operation on the image through an R1 density estimation network of an R2 pedestrian detection network; and integrating the pedestrian detection or density estimation result of each block to form a density map, and mapping the estimated density map onto the layer through perspective transformation so as to conveniently carry out accurate supervision on the global crowd condition.

Description

technical field [0001] The invention relates to a crowd counting and density estimation method, in particular to a deep network-based multi-strategy global crowd analysis method, which belongs to the technical field of machine vision and artificial intelligence. Background technique [0002] With exponential population growth and deepening urbanization, the number and frequency of large-scale gatherings have increased dramatically, such as scenic spot tourism on statutory holidays, sports games, political rallies, public exhibitions, etc. For better management, to ensure environmental safety and personal safety, it is necessary to analyze the crowd, and pedestrian detection and crowd counting are the current research focus. The current detection and counting methods mainly include: [0003] 1. Methods based on individual statistics [0004] Through the top view of the camera, the human head is detected to effectively resist occlusion; the human body is detected through the...

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): G06K9/00G06K9/62G06N3/04
CPCG06V20/53G06N3/045G06F18/214G06F18/24
Inventor 郑宏赵云峰姜寿林张莹
Owner 苏州平江历史街区保护整治有限责任公司
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