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

Pedestrian detection method oriented to automobile auxiliary driving on the basis of neural network

A technology for vehicle assistance and pedestrian detection, which is applied in the field of pedestrian detection, can solve problems such as complexity, and achieve the effect of high prediction accuracy and good generalization ability

Inactive Publication Date: 2018-11-30
SHANGHAI UNIV
View PDF5 Cites 2 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

In the ADAS-oriented far-infrared pedestrian detection system, the test scene may change greatly with factors such as climate, dynamic background, and many road traffic participants, and even tend to be complicated. Strong non-rigid body characteristics, its appearance mode is easily affected by factors such as motion posture, clothing diversification (such as reflectivity and transmittance affecting infrared thermal radiation), imaging scale, imaging angle of view, occlusion and complex background, etc. and the characteristics of the variety of variation between classes

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
  • Pedestrian detection method oriented to automobile auxiliary driving on the basis of neural network
  • Pedestrian detection method oriented to automobile auxiliary driving on the basis of neural network
  • Pedestrian detection method oriented to automobile auxiliary driving on the basis of neural network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0052] This embodiment introduces the process of the experimental car built by the present invention.

[0053] The present invention adopts a TMS320DM6437 EVM hardware platform with a main frequency of 600MHz, equipped with a monocular far-infrared sensor. The installation height is 24 cm (with a level road as a reference surface). The camera has a spatial resolution of 384 × 288 pixels, a data acquisition rate of 25 frames per second, a focal length of 3 meters, and a field of view of 28° × 21°.

[0054] The process of the present invention for collecting training and testing data includes:

[0055] S1. The driving speed of the data collection platform (car) should not exceed 60 km / h; the data collection locations mainly include suburban and semi-suburban scenes; the time period of data collection is from 7:30 to 9:30 in the evening; the collected data It is stored in lossless avi format, and the resolution of the image sequence obtained in this video format is 352 pixels×2...

Embodiment 2

[0057] This embodiment introduces the specific process of adopting the pedestrian detection method based on the neural network for assisted driving of automobiles.

[0058] In the present invention, a neural network adapting to requirements is first constructed, and then optimized according to different changes to achieve excellent detection effects, and comparison and calculation are performed.

[0059] When performing comparative observation and calculation, the present invention uses a neural network for pedestrian detection of the driving assistance system, learns pedestrian steps through iterative training, and uses detection accuracy and detection speed as effect comparisons.

[0060] The present invention compares the detection accuracy and speed effects produced by training and learning in different scenarios, including:

[0061] Table 1 Performance comparison of pedestrian detection in different scenarios

[0062] Scenes

[0063] Compared with the prior art...

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 detection method oriented to automobile auxiliary driving on the basis of a neural network. The method comprises the following steps that: A: putting forward a farinfrared pedestrian detection method based on probability template matching, and establishing a multiscale probability template according to the movement orientation of a pedestrian so as to alleviate the problem of a large pedestrian intraclass variance caused by an appearance pattern; B: from the perspective of ROIs (Regions Of Interest) extraction, utilizing image gradient information to carryout preliminary positioning on a vertical band-shaped image region which may contain the pedestrian; and C: putting forward a Boosting-style inductive transfer learning algorithm which presents ideally for the pedestrian detection method when a scene factor greatly changes. An optimization comparison method is adopted to obtain an optimal detection result, different detection networks and training mechanisms can be set according to practical requirement situations, flexibility is high, and dynamic performance is good. The invention belongs to a detection algorithm optimization method.

Description

technical field [0001] The invention belongs to a detection algorithm optimization method, and is a pedestrian detection method based on a neural network in a special scene. Background technique [0002] Pedestrian detection for ADAS requires real-time and accurate detection of pedestrians and their positions in front of the vehicle, and judgment of possible accident hazards, so as to take appropriate preventive measures. In the ADAS-oriented far-infrared pedestrian detection system, the test scene may change greatly with factors such as climate, dynamic background, and many road traffic participants, and even tend to be complicated. Strong non-rigid body characteristics, its appearance mode is easily affected by factors such as motion posture, clothing diversification (such as reflectivity and transmittance affecting infrared thermal radiation), imaging scale, imaging angle of view, occlusion and complex background, etc. and inter-category variation characteristics. Cont...

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): G06K9/00G06K9/62
CPCG06V40/10G06F18/241
Inventor 张姣周传宏
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