Computer Vision Based Approach for Traffic Violation Detection

Abstract

Abstract -The ever growing number of vehicles in a country present a variety of problems including but not limited to; infrastructural problems, air and water pollution and accidents with the latter being the most apparent. The main cause for this being traffic violations. This research was carried out with the intention of detecting motor traffic violations using CCTV footages. While there have been attempts to create automated traffic violation detection systems over the years, these studies have mostly been focused on more streamlined and sparse traffic conditions such as highways. But, the type of traffic conditions observed in Sri Lanka among other developing countries are unruly and chaotic. This paper proposes an automated real-time traffic violation detection system for highly congested and unruly road traffic conditions. The proposed system uses computer vision techniques, machine learning technology in creating a traffic violation detection system.