Research on heavy truck recognition algorithm based on deep learning Online publication date: Mon, 24-Oct-2022
by Huan Wang; Dun Zhang; Zhikai Huang
International Journal of Wireless and Mobile Computing (IJWMC), Vol. 23, No. 2, 2022
Abstract: The Chinese economy has developed rapidly by taking advantage of convenient and rapid freight transport. Heavy trucks have always attracted much attention as the leading force in freight transportation. Although heavy trucks have significant advantages in freight transportation, their high emissions that cause air pollution and high accident rates have always been criticised. Monitoring of heavy trucks has been improving in China, and this paper adopts deep learning to identify heavy trucks to strengthen their supervision. Given the multi-target recognition problem in the actual scene, this paper uses a one-stage algorithm for target detection. The representative network SSD (Single Shot Multi-Box Detector) and YOLO (You Only Live Once) are compared. YOLO adopts the YOLOv5s structure, and the SSD network is subdivided into two networks by replacing the backbone structure. One backbone structure is VGG, and the other is Mobilenetv2. The final experimental results show that the SSD network with VGG as the backbone structure achieves the best Map value of 93.24%, which is 6.02% and 8.14% higher than the SSD and YOLOv5s training models with Mobilenetv2 backbone structure, respectively.
Existing subscribers:
Go to Inderscience Online Journals to access the Full Text of this article.
If you are not a subscriber and you just want to read the full contents of this article, buy online access here.Complimentary Subscribers, Editors or Members of the Editorial Board of the International Journal of Wireless and Mobile Computing (IJWMC):
Login with your Inderscience username and password:
Want to subscribe?
A subscription gives you complete access to all articles in the current issue, as well as to all articles in the previous three years (where applicable). See our Orders page to subscribe.
If you still need assistance, please email subs@inderscience.com