Intelligent vehicle license plate recognition by deploying deep learning model for smart cities

Abstract

According to surveys, almost 3.5 lakh road accidents have taken place in India within a year in 2020. Almost 1.4 lakh people have died due to road accidents. The major cause of road accidents in India is over speeding. The population of India makes it difficult for the police to monitor every single vehicle that breaches the traffic rules. Many solutions have been proposed to maintain traffic and make the drivers and pedestrians follow the traffic rules, especially the speed limit. Yet, the drivers don’t seem to obey the rules and stay within the speed limit at the places which were not monitored by police or other humans. To resolve this issue, this research aims the development of a deep learning model which is capable of predicting the registration number of a certain vehicle in the road and this reach helps to turn the normal city into a smart city. By applying the model in real life, the vehicles which involve in traffic crimes a repeated number of times can be identified and the driving license of the driver can be blocked. This deep learning model is constructed using the convolutional neural network theory and is very easy to test and train. The model is then trained again and again to reach a higher precision. The efficiency of the model is also tested sometimes to ensure the efficiency is high. The efficiency of the model is monitored using three parameters named recall, precision, mean average precision.