Road Traffic Vehicle Detection Method Using Lightweight YOLOv5 and Attention Mechanism

With the rise in the number of vehicles on the streets, urban road problems are becoming more and more prominent. As the vehicle of the road subject, it is the subject of the problem of intelligent transportation system. In this paper, we discuss the vehicle detection problem in intelligent transpor...

Full description

Saved in:
Bibliographic Details
Published in2022 7th International Conference on Image, Vision and Computing (ICIVC) pp. 201 - 207
Main Authors Wang, Yunzhen, Ma, Hongbing, Li, Liangliang
Format Conference Proceeding
LanguageEnglish
Published IEEE 26.07.2022
Subjects
Computational modeling
Convolution
Data models
depthwise separable convolution
Intelligent transportation systems
Object detection
Roads
SENet
Testing
vehicle model identification
YOLOv5
Online AccessGet full text

Cover

More Information
Summary:With the rise in the number of vehicles on the streets, urban road problems are becoming more and more prominent. As the vehicle of the road subject, it is the subject of the problem of intelligent transportation system. In this paper, we discuss the vehicle detection problem in intelligent transportation system, and propose a lightweight YOLOv5 model combining SENet attention mechanism and depthwise separable convolution for the complex situation of vehicles in street video, such as vehicle occlusion, vehicle misdetection and vehicle omission. The feasibility and effectiveness of the method are verified by testing on UA-DETRAC and self-built vehicle dataset.
DOI:10.1109/ICIVC55077.2022.9886525