JIITA, Vol.9 No.3 pp.1130-1136 (2025), DOI: 10.22664/ISITA.2025.9.3.1130
Sang Suh, and Bilal Mushtaq
Abstract. Overspeeding is a major cause of fatal accidents and car crashes around the world. Drivers that aren’t timely held accountable
for their recklessness become a threat for themselves and for those sharing the road with them. Numerous researchers have presented
machine learning and deep learning-based object detection and recognition techniques that aim to accurately
detect vehicles but have come across hurdles such as varying illumination and weather conditions, overlapping vehicle images
in complex traffic situations, the need for powerful GPU based computers at surveillance, and slow detection speeds.
The proposed work VRSMY9 (Vehicle Recognition and Speed Monitoring System using YOLOv9)
powered by YOLOv9 and EasyOCR focuses on enhancing the overall detection speed in real-time without compromising too
much on the detection accuracy whilst being lightweight enough for computing devices that are deprived of powerful GPUs.
The system detects vehicles, estimates their travelling speeds, detects the license plates of the overspeeding vehicles
resulting in license plate extraction of all the violators, and then keeps a stored record of them.
Results show that the system is highly precise and has decent accuracy with a high detection speed but isn’t accurate enough when
handling challenging images. The scope of this work is extended but not limited to autonomous vehicles, unmanned aerial vehicles,
intelligent transportation systems, robotics, intelligent AI agents, domestic security, healthcare and all other areas where high speed
accurate detection is needed on an embedded smart device that may not be computationally super powerful.
Keywords; YOLOv9, EasyOCR, Vehicle Detection, License Plate Recognition, Speed Estimation, Real-time Traffic Surveillance
Fullpaper: