Helmet Detection and Automatic License Plate Recognition: A Simple Safety-Based Approach

Abstract

Road safetyfor two-wheeler riders remains a significant challenge, particularly since many people continue to ride without helmets. Manual monitoring of such violations is challenging for traffic authorities because it requires constant attention and a large amount of manpower. To address this problem, this research presents an automated system that combines helmet detection with automatic license plate recognition using modern machine-learning techniques. The system uses deep-learning models, such as YOLO, to identify whether a rider is wearing a helmet by analysing images or video frames in real time. If a rider is found without a helmet, the system automatically detects the motorcycle’s license plate and reads the characters using an OCR-based approach. The proposed method reduces the need for human supervision and increases the speed and accuracy of violation detection. The system is designed to work with CCTV footage and can handle various real-world challenges such as different lighting conditions, camera angles, and backgrounds. By integrating helmet detection and license plate recognition into a single pipeline, this research demonstrates a practical solution for traffic monitoring and enforcement. The work demonstrates that AI-powered systems can play a crucial role in building safer roads, enhancing rule compliance, and supporting smart-city initiatives.

Introduction

Road accidents involving two-wheelers continue to rise mainly because many riders do not wear helmets. Manual checking by traffic police is difficult, slow, and often inaccurate, especially in crowded areas. To solve this problem, modern computer-vision systems can automatically detect riders without helmets and identify their vehicle number plates using deep learning.

Shift From Traditional Methods to Deep Learning

Early approaches such as Haar cascades, HOG features, SVMs, and edge detection worked only on simple images and failed in real traffic due to poor lighting, shadows, and motion blur.
With the growth of deep learning, advanced models like Faster R-CNN, SSD, and especially YOLO significantly improved real-time helmet detection and object recognition.
Recent versions such as YOLOv7, YOLOv8, and modern OCR models using CNNs and Transformers now offer high accuracy even in complex scenes and low-quality video.

Integrated Helmet Detection + ALPR System

The system first detects whether the rider is wearing a helmet using a trained YOLO model. If a violation is found, a second model detects and crops the vehicle’s number plate. Optical Character Recognition (OCR)—using tools like Tesseract or EasyOCR—reads the text on the plate.
The system stores the violation frame, plate number, date, and time, enabling automated challan generation without manual intervention.

Findings from Literature Review

• Traditional algorithms worked only on controlled images.

• Deep learning (2016–2020) improved accuracy but struggled with very small helmets and unclear plates.

• Advanced models (2020–2024) like YOLOv4–v8 and CRNN/Transformer OCR increased robustness and real-time performance.

• State-of-the-art systems integrate cloud computing, edge devices, and smart-city platforms for automated monitoring.

Methodology

Images and videos of two-wheeler riders are collected and labelled. A YOLO model is trained to detect helmets and non-helmet cases. A second model identifies the number plate, followed by OCR to extract plate text. All details are saved for enforcement.

Helmet Detection

Modern deep learning models accurately detect helmets in real time and perform well even under challenges like low light, motion blur, and different helmet designs. Techniques such as data augmentation and attention networks improve performance.

Automatic License Plate Recognition (ALPR)

ALPR includes plate detection, character extraction, and text recognition. Preprocessing methods—grayscale, blur, thresholding, CLAHE, sharpening—are used to enhance OCR accuracy.
Among all methods, CLAHE gives the highest OCR confidence, while adaptive thresholding performs the worst.

Real-Time Implementation Using YOLOv8

YOLOv8 provides high accuracy, anchor-free detection, and excellent real-time speed. It works with:

• Standard webcams

• Mobile/IP cameras

• Edge devices

Its improved architecture helps detect small helmets and plates more reliably.

Future Scope

• Training with bigger, more diverse datasets

• Better night-time and low-resolution performance

• Integration with government databases for automatic challans

• Deployment on roadside cameras using lightweight models

• Expansion to other violations (speeding, triple riding, signal jumping)

Smart City Integration

The system can be added to smart surveillance networks with IoT devices and cloud platforms to provide city-wide automated monitoring.

Multi-Violation Detection

The same system can be extended to detect:

• Triple riding

• Wrong-side driving

• Mobile phone usage

• Overspeeding

This makes it suitable for comprehensive traffic-rule enforcement.

Conclusion

This project shows how modern technology can help improve road safety by automatically detecting helmet violations and identifying the vehicles involved. By using deep learning models, the system can quickly check whether a rider is wearing a helmet and then read the motorcycle’s license plate if a violation occurs. This reduces the need for manual monitoring and makes the process more accurate and efficient. The combination of helmet detection and license plate recognition proves to be a practical solution for real-world traffic environments. Even though challenges like poor lighting, unclear number plates, and low-resolution video still exist, the results show that AI-based systems can perform well with proper training and good-quality datasets. Overall, this work demonstrates that automating traffic rule enforcement is not only possible but also highly beneficial. With further improvements in model accuracy and better camera setups, such systems can be used in smart cities to support traffic police, reduce accidents, and encourage safer riding behaviour.

References

[1] Viola and Jones, “Robust Real-time Object Detection\", ICCV2001. [2] Navneet Dalal and Bill Triggs, “Histogram of oriented gradients for human detection\". [3] Ross, Jeff, Trevor, and Jjitendra\"Rich feature Hierarchy for Accurate object Detection\". [4] Shaoqing Ren, Kaiming He, Ross Girshick, Jian Sun, \"Fast R-CNN\" (Submitted on 4 Jun 2015 (v1), last revised 6 Jan 2016 (this version, v3)). [5] Joseph Redmon, Ali Farhadi, \"YOL09000: Better, Faster, Stronger\", University of Washington, Allen Institute Of AI.[6] Joseph Redmon, Ali Farhadi, \"YOLOv3: An Incremental Improvement\", University of Washington, Allen Institute of AI. [6] Wei Liu, Dragomir Anguelov, Dumitru Erhan, Christian Szegedy, Scott Reed, Cheng-Yang Fu, Alexander C. Berg, \"SSD: Single Shot MultiBox Detector\". [7] A. Adam, E. Rivlin, I. Shimshoni, and D. Reinitz, “Robustreal-time unusual event detection using multiple fixed-location monitors, IEEE Transactions on Pattern Analysis and Machine Intelligence, vol.30, no. 3, pp.555-560, March 2008. [8] AlexeyAB, https://github.com/AlexeyAB/darknet#requirements. [9] C.Y. Wen, S.-H.Chiu, J-I.Liaw, and C.-P. Lu, “The safety helmet detection for atm\'s surveillance system via the modified Hough transform, ”in IEEE 37th Annual International Carnahan Conference on Security Technology, 2003, pp.364-369. [10] C.C. Chiu, M.Y. Ku, and H.T. Chen, “Motorcycle detection and tracking system with occlusion segmentation, ”in WIAMIS \'07, USA, 2007 [11] A. Hirota, N. H. Tiep, L. Van Khanh, and N. Oka, Classifying Helmeted and Non-helmeted Motorcyclists.Cham: Springer International Publishing, 2017, pp.81-86 [12] Lokesh Allamki, Manjunath Panchakshari, Ashish Sateesha, and K. S. Pratheek, “Helmet Detection using Machine Learning and Automatic License Plate Recognition”, IRJET, Vol. 6, Issue 12, 2019.

Copyright

Copyright © 2025 Bithi Roy. 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.