Eureka AIR delivers breakthrough ideas for toughest innovation challenges, trusted by R&D personnel around the world.

A Real-time Detection Method of Pedestrians and Vehicles Based on Lightweight Deep Network

A deep network, real-time detection technology, applied in the field of deep learning technology, can solve the problems of low accuracy, large number of parameters, too large number of parameters, etc., to achieve good practicability and real-time performance, small number of model parameters, accurate detection high degree of effect

Active Publication Date: 2021-02-05
UNIV OF ELECTRONICS SCI & TECH OF CHINA
View PDF4 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0007] (3) The real background of pedestrians and vehicles is complicated;
[0010] (6) The occlusion situation of pedestrians and vehicles is complex and diverse;
However, these two methods have certain limitations in practical applications. The accuracy of the former is not high in practical applications. The ability to extract features can improve the detection accuracy of the deep network, but at the same time, the model parameters and computational complexity of the network are also greatly increased, which undoubtedly puts forward certain requirements for the memory and computing power of the hardware platform. The complexity is too large, especially on some embedded platforms with limited memory size and computing power, it is difficult to achieve real-time effects
[0015] VGG-16 is an existing classification network with good effect. It is often used as the feature extraction part of the target detection network. style platform

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 Real-time Detection Method of Pedestrians and Vehicles Based on Lightweight Deep Network
  • A Real-time Detection Method of Pedestrians and Vehicles Based on Lightweight Deep Network
  • A Real-time Detection Method of Pedestrians and Vehicles Based on Lightweight Deep Network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0030] The present invention proposes a new convolutional neural network (CNN) structure to complete the feature extraction part of pedestrians and vehicles. The specific network structure is as figure 1 Shown:

[0031] The network consists of 8 layers in total (5 meta-modules + 3 convolutions). Different from the linear structure of VGG-16, this network adopts a skip connection method between layers to fuse the shallower features of the network with deeper features: the first convolutional layer outputs the feature spectrum to the first meta-module , the convolutional layer extracts features from the image through a series of filters, and adjusts the parameters through regularization and activation operations to make the network converge better. The first meta-module outputs the feature spectrum to the second meta-module, and the first convolutional layer The output feature spectrum is fused with the output feature spectrum of the second meta-module and then output to the t...

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 provides a real-time detection method for pedestrians and vehicles based on a lightweight deep network. The lightweight deep network uses 5 modules + 3 kinds of convolution operations, and the meta-module only contains 2 kinds of convolution operations to realize the feature extraction function. . The jump connection mode between different modules and the more robust feature spectrum fusion technology of the present invention enable the network to achieve a better detection effect on pedestrians and vehicles when the model parameters are small, and can effectively detect images or Pedestrian vehicles in the video. The new deep network proposed by the present invention has the advantages of small model parameters, small computational complexity and high detection accuracy, and can realize real-time detection of pedestrians and vehicles on the embedded platform, and has good practicability and real-time performance .

Description

technical field [0001] The invention relates to deep learning technology in image processing. Background technique [0002] With the continuous growth of urban economic level and population, the number of vehicles and pedestrians on traffic roads also increases accordingly. [0003] The ensuing series of traffic problems such as road traffic congestion and frequent traffic accidents have put forward higher requirements for urban traffic construction. Therefore, a more efficient and rapid real-time detection method for pedestrians and vehicles has broad application prospects and urgent market demand. [0004] The research on pedestrian and vehicle detection methods at home and abroad has been carried out for decades. During this period, many outstanding scholars have proposed many effective detection methods. However, as far as the ultimate goal of computer vision technology-equivalent to the recognition ability of human beings, there is still a certain gap in the detection...

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 Patents(China)
IPC IPC(8): G06N3/04G06N3/08G06K9/62
Inventor 李宏亮孙玲张文海翁爽董蒙
Owner UNIV OF ELECTRONICS SCI & TECH OF CHINA
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
Eureka Blog
Learn More
PatSnap group products