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

Pedestrian target tracking method based on depth learning

A pedestrian target and deep learning technology, applied in image analysis, image enhancement, instruments, etc., can solve the problems of training sample pollution, feature pertinence is not strong, and cannot be directly applied to usage scenarios, so as to improve accuracy and efficiency Effect

Active Publication Date: 2019-01-04
HUAZHONG UNIV OF SCI & TECH
View PDF3 Cites 24 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0008] Aiming at the defects of the prior art, the purpose of the present invention is to solve the problems of the existing pedestrian target tracking method that the extracted features are not highly targeted, the accuracy of the deep convolution feature tracking target position is not high, and the occlusion brings pollution of the training samples, etc. Technical problems that cannot be directly applied to actual usage scenarios

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 target tracking method based on depth learning
  • Pedestrian target tracking method based on depth learning
  • Pedestrian target tracking method based on depth learning

Examples

Experimental program
Comparison scheme
Effect test

example

[0135] In order to prove that the pedestrian target tracking method based on deep learning has advantages in performance and efficiency, the present invention conducts verification and analysis through the following experiments.

[0136] A. Experimental data

[0137] The present invention conducts experiments on the MOT-16 data set, which contains 14 video sequences.

[0138] B. Experimental platform

[0139] Hardware: CPU Intel Xeon E5-2650v3, memory 64G, GPU GeForce GTX TITANX, video memory 12G, hard disk 4TB 7200 rpm.

[0140] Software: operating system windows8, Ubuntu16.04, experimental platform Caffe, MatconvNet, Matlab.

[0141] C Pedestrian Object Tracking Evaluation Criteria

[0142] Mean Overlap Precision (mOP), Speed ​​Evaluation Standard FPS, Average Tracking Time.

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 target tracking method based on depth learning, which combines depth learning and correlation filtering to track the target, and effectively improves the accuracyof tracking on the premise of ensuring real-time tracking. Aiming at the problem of large change of target posture in tracking process, the deep convolution feature based on pedestrian attribute is applied to tracking. Aiming at occlusion problem, cosine similarity method is used to judge occlusion in order to avoid the introduction of dirty data caused by occlusion. In order to improve the efficiency and solve the problem of using deep convolution features in correlation filters, a bilinear interpolation method is proposed to eliminate quantization errors and avoid repeated feature extraction, which greatly improves the efficiency. Aiming at the problem of high-speed motion of target, a preselection algorithm is proposed, which can not only search the global image, but also can be used asa strong negative sample to join the training, so as to improve the distinguishing ability of the correlation filter.

Description

technical field [0001] The present invention belongs to the field of computer vision, and more specifically, relates to a pedestrian target tracking method based on deep learning. Background technique [0002] Recently, terrorist incidents at home and abroad have occurred frequently, seriously threatening the safety of people's lives and property, as well as their healthy and happy lives. With the construction of "Safe City", the construction of a large number of cameras not only guarantees the safety of people's lives and property, but also brings about the explosive growth of video data, bringing video analysis and processing technology. A great challenge has come. Most of the large amount of data generated by video surveillance systems are related to people, and most of what we care about are people's characteristics, postures, actions, behaviors, etc. Therefore, it is very important to obtain information related to people. Pedestrian target tracking is to distinguish p...

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): G06T7/246G06K9/62
CPCG06T7/251G06T2207/20081G06T2207/20084G06T2207/30196G06T2207/10016G06F18/241G06F18/214
Inventor 凌贺飞余成跃李平
Owner HUAZHONG UNIV OF SCI & TECH
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