Detection of Non Helmet Riders and Extraction of Licence Plate Number

The current scenario of India's traffic laws poses a number of challenges, each of which may be addressed in a variety of ways. An increasing incidence of motorbike and moped-related fatalities have been linked to the common practice of driving without a headgear in India. The current system relies heavily on CCTV records to keep tabs on traffic offences, forcing cops to manually examine individual frames and, if the rider isn't wearing a helmet, focus in on their numberplate. However, this calls for a significant investment of time and resources, since traffic offences are common and the number of motorcyclists grows daily. What if there were a technology that could detect whether or not a motorcyclist was breaking the law by failing to wear a helmet and, if so, get the vehicle's registration number from the database? This study has been accomplished with recent success using HaaR, HoG, LBP, R-CNN, CNN features, etc. However, the pace, precision, and effectiveness of object identification and categorization are severely constrained by these works. In an effort to automate the process of finding instances of the traffic offense of not donning a headgear and retrieving the numberplate of vehicles, this study develops a Non-Helmet Driver identification method. The core idea is to use three tiers of DL for Image Classification. Individual, motor bike, headgear, and licence plate are all recognisedutilising YOLOv2's increasing levels of sophistication. Then, OCR is used to get the licence plate number. Particularly for the extraction of licence plate numbers, all of these methods are governed by a set of predetermined parameters and limits. Since moving images are used as input, the processing time is critical. By combining the aforementioned methods, we have developed a comprehensive system for identifying helmets and extracting licence plate numbers.