A Pedestrian Detection Method Combining Multiple Models and Multiple Thresholds

A pedestrian detection and multi-threshold technology, which is applied in character and pattern recognition, instruments, computer components, etc., can solve the problem of pedestrian detection in the image, and achieve the effect of improving detection efficiency

Active Publication Date: 2017-08-08
谭家湾遗址公园运营管理(桐乡乌镇)有限公司
View PDF4 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0012] In order to solve the technical problem that the current single detection method cannot effectively and accurately detect pedestrians in the image, the present invention provides a pedestrian detection method that integrates multiple detection methods and means to achieve accurate pedestrian detection with a combination of multiple models and multiple thresholds

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 Pedestrian Detection Method Combining Multiple Models and Multiple Thresholds
  • A Pedestrian Detection Method Combining Multiple Models and Multiple Thresholds
  • A Pedestrian Detection Method Combining Multiple Models and Multiple Thresholds

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0046] The voc2007 mentioned in this embodiment comes from: The PASCAL Visual Object ClassesChallenge 2007, see:

[0047] http: / / pascallin.ecs.soton.ac.uk / challenges / VOC / voc2007 / ;

[0048] inria, caltech, tud and eth from: "Related Datasets" of "Caltech PedestrianDetection Benchmark"; see:

[0049] http: / / www.vision.caltech.edu / Image_Datasets / CaltechPedestrians /

[0050] In this embodiment, the algorithm principle of the "DPM detection sub-module" is detailed in the following papers:

[0051] Object Detection with Discriminatively Trained Part Based Models, P. Felzenszwalb, R. Girshick, 2010;

[0052] For the algorithm program, see: http: / / www.cs.berkeley.edu / ~rbg / latent / ; the DPM detection sub-module in this embodiment is directly from the above open source program without modification.

[0053] see Figure 7 , the DPM classifier uses a combination of basic SVM and struct-Latent-SVM, uses a sliding window of a certain size, moves on the image with a certain step size, and...

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 combined with multiple models and multiple thresholds. By combining two different detection methods, different detection models and detection thresholds are selected to form multiple detection sub-modules to form a parallel detection structure and simultaneously detect one image. The detection results form a collection as a candidate set; then remove false positives and merge the same detections according to the reliability method. The average detection rate has been increased by about 20%, reaching more than 85%, and the average false positive rate is <10%. Compared with the detection by DPM or ICF alone, if the detection rate reaches 85%, the average false positive rate is >30%. In terms of detection efficiency, the detection efficiency has been greatly improved.

Description

technical field [0001] The invention relates to a pedestrian detection method combined with multiple models and multiple thresholds. Background technique [0002] At present, for still images, there are mainly two better pedestrian detection methods: [0003] Deformable Part Model (DPM: Deformable Part Model) [0004] Object Detection with Discriminatively Trained Part Based Models, P. Felzenszwalb, R. Girshick, 2010; [0005] Cascade Object Detection with Deformable Part Models, P. Felzenszwalb, R. Girshick, 2010. [0006] Integrated Channel Features Model (ICF: Integral Channel Features) [0007] Pedestrian Detection: An Evaluation of the State of the Art, Piotr Dollar, 2012; [0008] The Fastest Pedestrian Detector in the West, Piotr Doll ar, 2010; [0009] Integral Channel Features, Piotr Dollar, 2009. [0010] These methods, for public typical pedestrian databases, can often achieve better detection results, which may be because the samples in these pedestrian dat...

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): G06K9/66
CPCG06V40/103
Inventor 徐晓晖
Owner 谭家湾遗址公园运营管理(桐乡乌镇)有限公司
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