This paper focuses on vehicle detection at highway toll gates. The study involves the detection of vehicles within a custom dataset, which comprises four vehicle classes: car, truck, bus, and motorcycle. The YOLOv4 model is utilized for training and is complemented by the implementation of the Centroid tracking algorithm. In the proposed vehicle track dataset, each class comprises 1500 annotated images for training and 300 images in the testset. These images are sourced from the Google Open Images V6 dataset, and the OIDv4 toolkit is employed to scrape dataset images. The primary objective of this study is to train a vehicle detection model and evaluate the effectiveness of various tracking algorithms, with a specific focus on the centroid tracker for real-time application at the toll gates. Importantly, this model is designed to operate efficiently on conventional and cost-effective GPUs and demonstrates a commendable level of precision.
Light Weighted YOLO Vehicle Tracker for Toll Gate Using Centroid Algorithm
Lect. Notes Electrical Eng.
International Conference on Emerging Electronics and Automation ; 2023 ; Silchar, India December 15, 2023 - December 17, 2023
2025-02-13
12 pages
Article/Chapter (Book)
Electronic Resource
English
An Improved Weighted Centroid Location Algorithm
British Library Online Contents | 2010
|A Real-Time Wrong-Way Vehicle Detection Based on YOLO and Centroid Tracking
ArXiv | 2022
|Noise effects on centroid tracker aim point estimation
IEEE | 1988
|Weighted Centroid Localization Algorithm Based on Maximum Likelihood Estimation
British Library Online Contents | 2016
|