Abstract:

Motorcycles have traditionally served as the primary mode of transportation in various countries across the globe. However, there has been a rise in motorcycle-related incidents resulting in harm or fatalities over time. Inadequate head protection remains one of the primary factors leading to lethal outcomes resulting from such incidents. This research endeavours to tackle the issue by designing a system capable of identifying motorcyclists who are not wearing helmets instantly. The proposed system aims to facilitate the use of helmets among motorcycle riders in real-time, helping reduce fatalities due to head injuries, which are a major concern in many countries. Additionally, the system also includes a method for real-time number plate detection and recognition, which has numerous good significances in law enforcement and surveillance. This helps the officers to keep on the people who violate traffic rules and regulations, ensuring their safety as well as that of pedestrians. The system utilizes state-of-the-art deep learning algorithms, such as YOLOv3, to achieve real-time helmet-wearing detection with unparalleled speed and accuracy of 99%.

Keywords: Machine learning algorithm Yolo ,dataset ,helmet detection, license plate

Introduction:

Due to the large number of vehicles in circulation, studies of intelligent traffic systems have increased. The majority of studies focus on the detection, recognition, tracking and counting of vehicles and on the estimation of traffic parameters. Motorcycles are a predominant method of transport in many countries. The main advantages of motorcycles are their low price and operation cost compared with other vehicles. According to DENATRAN(National Department of Traffic), Brazil maintained a fleet of 20.281,986 motorcycles in 2013 .
The number of accidents involving motorcycles has increased during the last decade. According to the DNIT, a total of 34,635 motorcycles were involved in accidents in Brazil in 2011 . According to a study of traffic accidents , 14,666 traffic fatalities occurred in 2011.
This global problem is prevalent. According to statistics from 2012 , more than 17 million motorcycles were officially registered in Thailand in 2010. This large number of vehicles increased the potential for a large number of traffic accidents. According to a report on road safety published by the World Health Organization (WHO) in 2013 , the estimated rate of deaths per 100.000 people in Thailand was 38.1. According to the report, Thailand has the third highest traffic mortality rate in the world. In Brazil, the traffic mortality rate was 22.5; in the United States (US), the rate was 11.4. The report noted that approximately 25 % of traffic fatalities in Brazil involved motorcyclists. The main safety equipment used by motorcyclists is the helmet. Although helmet use is mandatory, some motorcyclists do not use them or incorrectly use them. According to the US Department of Transportation’s National Highway Traffic Safety Administration (NHTSA), only 66 % of motorcyclists used a helmet in accordance with the law . The US Centers for Disease Control and Prevention presented a Morbidity and Mortality Weekly Report (MMWR) in 2012, in which the results indicated that 12 % of fatally injured motorcyclists were not using a helmet in states with universal helmet laws. In the states that adopt partial helmet laws , the number increased to 64 %. A total of 79 % of motorcyclists use helmets in states without helmet laws

The motivation of this work is to improve surveillance on the main roads in locations where the use of helmets is mandatory. These data reveal the need for increased enforcement. of traffic laws, particularly for offenses for which there are no automatic detection methods.
The increase in the number of motorcyclists using helmets causes a decrease in the number of accidents with victims, which is high in those countries. This study aimed to develop a computational vision methodology to detect motorcyclists without helmets. A two-stage strategy was developed, namely, the detection of motorcycles and the detection of helmet use. The main contribution of this work is related to a general solution proposed to detect motorcyclists without helmets, not the development of new algorithms for each stage of the system. The proposed methodology has been designed to be applied in American and Asian countries such as Brazil, USA, India, Thailand and others. In these countries the number of motorcyclists and accidents are high, as shown in the statistics. Moreover, in some of those countries there are laws that require helmet use by motorcyclists.

Problem Statement:

Increasing Injuries and Fatalities: The absence of helmets significantly raises the risk of head injuries in accidents. The lack of automatic detection can decrease helmet usage, thereby increasing injuries and fatalities.

Objective:

The objective of the Helmet and Number Plate Detection and Recognition project using YOLO (You Only Look Once) is to develop a robust and efficient system capable of identifying whether motorcyclists are wearing helmets and detecting and recognizing vehicle number plates in real-time. This system aims to enhance road safety and law enforcement by automating the process of monitoring helmet compliance and identifying vehicles.