Open AccessEhicle tracking based on video in fast traffic scenarios
Author(s) -
Liao Ying Zheng,
Gang Li,
Xinyue Zhang
Publication year - 2019
Publication title -
iop conference series. materials science and engineering
Language(s) - English
Resource type - Journals
eISSN - 1757-899X
pISSN - 1757-8981
DOI - 10.1088/1757-899x/677/4/042118
Subject(s) - computer science , redundancy (engineering) , tracking (education) , artificial intelligence , convolution (computer science) , computer vision , video tracking , real time computing , video processing , psychology , pedagogy , artificial neural network , operating system
This paper is aiming at the problem of low tracking accuracy of moving vehicles in fast traffic scenarios, the ECO algorithm based on correlation filtering is applied to real-time tracking to achieve stable and accurate tracking of target vehicles. Through the factorized convolution operator, the extracted features are more comprehensive and efficient; the optimization of the training set effectively reduces redundancy; and the more representative correlation filter is used to prevent over-fitting; a simple update model is used to prevent model drift. The experimental results show that the proposed method finally achieved a tracking effect of 97% recognition rate.