SafeRide AI: Helmet and Number Plate Detection

Abstract

The rapid increase in two-wheeler usage has led to a significant rise in traffic accidents, many of which result in fatalities due to the non-compliance of safety regulations such as wearing helmets. This paper presents the design and implementation of SafeRide AI, a real-time computer vision framework designed to automate the detection of motorcyclists not wearing helmets and the subsequent extraction of their vehicle license plate numbers. By leveraging the YOLOv8 (You Only Look Once) architecture for high-speed object detection and EasyOCR for robust Optical Character Recognition (OCR), SafeRide AI provides a seamless, end-to-end pipeline for traffic enforcement. The system utilizes a specialized dataset to distinguish between \"with helmet\" and \"without helmet\" classes while simultaneously locating number plates. A sophisticated association logic is implemented to link violations directly to the corresponding vehicle identifiers. Experimental results indicate that our implementation achieves a high Mean Average Precision (mAP) for detection while maintaining real-time processing speeds of up to 30 frames per second on standard GPU hardware. This research contributes a scalable solution for urban traffic monitoring and the promotion of road safety standards through automated AI intervention.

Introduction

Road traffic safety is a major global issue, especially for motorcyclists, who are highly vulnerable to fatal head injuries due to lack of helmet use. Manual traffic monitoring is inefficient, prompting the need for automated solutions using Deep Learning (DL) and Computer Vision (CV).

SafeRide AI is an advanced system designed to detect helmet violations and identify offenders in real time. It uses a YOLOv8-based model for object detection and integrates license plate recognition within the same pipeline, overcoming limitations of earlier fragmented systems.

The system employs a multi-stage process: detecting riders, helmets, and number plates; associating them spatially to identify violations; and extracting license plate text using OCR (EasyOCR). It is supported by a user-friendly Streamlit dashboard for analyzing images and video streams.

Compared to manual monitoring and older models, SafeRide AI achieves higher accuracy (89.2%), faster processing (28–35 FPS), and better violator traceability through automated license plate recognition. It performs well even in complex conditions and can be deployed efficiently, making it a practical solution for modern traffic enforcement.

Conclusion

SafeRide AI successfully demonstrates the viability of utilizing state-of-the-art computer vision to enhance public safety. By consolidating object detection and OCR into a unified, real-time pipeline, the platform effectively automates the enforcement of helmet compliance. The implementation of YOLOv8 ensures that the system is both fast enough for live traffic and accurate enough for legal accountability.

References

[1] Vaswani et al., \"Attention is all you need,\" in NeurIPS, vol. 30, 2017. [2] J. Redmon, S. Divvala, R. Girshick, and A. Farhadi, \"You Only Look Once: Unified, Real-Time Object Detection,\" CVPR, 2016. [3] Ultralytics, \"YOLOv8: Real-Time Object Detection and Image Segmentation,\" 2023. [Online]. Available: [4] J. Hou, \"EasyOCR: Ready-to-use OCR with 80+ supported languages,\" 2024. [Online]. Available: https://github.com/JaidedAI/EasyOCR [5] S. Zhang et al., \"Research on Helmet Detection and Identification Based on Improved YOLO,\" IEEE Access, 2022. [6] Google, \"PyTorch: Tensors and Dynamic neural networks in Python with strong GPU acceleration,\" 2024. [7] Streamlit Team, \"Building data apps with Streamlit,\" Documentation, 2024. [8] M. Sandler et al., \"MobileNetV2: Inverted Residuals and Linear Bottlenecks,\" CVPR, 2018. [9] K. He, X. Zhang, S. Ren, and J. Sun, \"Deep Residual Learning for Image Recognition,\" CVPR, 2016. [10] R. Girshick, \"Fast R-CNN,\" ICCV, 2015.

Copyright

Copyright © 2026 Anjala Pathan. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.