Mathematical Models Using Artificial Intelligence for Surveillance Systems

3Machine Learning and Imaging-Based Vehicle Classification for Traffic Monitoring Systems

Parthiban K.* and Eshan Ratnesh Srivastava

Vellore Institute of Technology, Vellore, Tamil Nadu, India

Abstract

A system of traffic monitoring plays a vital role in controlling the flow of traffic and ensuring road safety. A key element for traffic monitoring is the proper vehicle classification. It can be difficult to use traditional methods, which depend on manual observations or standard threshold procedures. Technologies that use machine learning and imaging provide an encouraging answer to this problem. With the use of machine learning and quantitative evaluation techniques, this research suggests a novel way for categorizing cars. Images of automobiles as well as data on their dimensions and other crucial details are taken using a variety of cameras. The images are subsequently analyzed by the deep convolutional neural network (Resnet), which then classifies the automobiles based on their characteristics (length, width, number of wheels and other visible dimensional attributes). When compared to benchmarks of conventional approaches, this strategy offers greater accuracy and dependability than those old methods. It is improved by the capability to take into account a variety of environmental influences, such as variations in lighting or weather. It is quick, effective and automatically process huge amounts of data because of the deep CNN. This approach is also adaptable to different ...