Resumen

PRAKASH-BORAH, Jyoti et al. Real-Time Helmet Detection and Number Plate Extraction Using Computer Vision. Comp. y Sist. [online]. 2024, vol.28, n.1, pp.41-53.  Epub 10-Jun-2024. ISSN 2007-9737.  https://doi.org/10.13053/cys-28-1-4906.

In the contemporary landscape, two-wheelers have emerged as the predominant mode of transportation, despite their inherent risk due to limited protection. Disturbing data from 2020 reveals a daily toll of 304 lives lost in India in road accidents involving two-wheeler riders without helmets, emphasizing the urgent need for safety measures. Recognizing the crucial role of helmets in mitigating risks, governments have made riding without one a punishable offense, employing manual strategies for enforcement with limitations in speed and weather conditions. In today’s world of advancing technology, we can leverage the power of computer vision and deep learning to tackle this problem. This can eliminate the need for constant human surveillance to be kept on riders and can automate this process, thus enforcing law and order as well as making this process efficient. Our proposed solution utilizes video surveillance and the YOLOv8 deep learning model for automatic helmet detection. The system employs pure machine learning to identify helmet types with minimal computation cost by utilizing various image processing algorithms. Once the helmet-less person is detected, the number plate corresponding to the rider’s motorcycle is also detected and extracted using computer vision techniques. This number plate is then stored in a database thus allowing further intervention to be done in this matter by the authorities to ensure penalties and enforce safety rules properly. The model developed achieves an overall accuracy score of 93.6% on the testing data, thus showcasing good results on diverse datasets.

Palabras llave : Image dataset; YOLOv8; deep learning model; object detection; image processing algorithms.