Image Based Automatic Traffic Surveillance System Through Number-Plate identification And Accident Detection

Abstract

The unpredictability and blockage of current transportation frameworks frequently produce traffic circumstances that endanger the security of the individuals in question. A simple invention can make easy to control the traffic system. This paper presents a programmed traffic observation framework to gauge significant traffic boundaries from video arrangements utilizing just captures from cameras. A traffic control kit is developed to detect over speeding cars on highways, number plates in Bengali, and initiating emergency call to 999 on detecting accidents. A GPS enabled traffic surveillance camera can detect the location of accident and send message to the traffic control room with location information of accident. A Python program is used to detect over speed which provides accurate speed of a vehicle very fast. OCR Tesseract is used to detect number plate which has very high performance in detecting noisy texts. To identify a case of accident, a simple Python code with Dens-net Architecture is used. A GSM module of the experimental kit initiate the call and message after analyzing the data through a code of C language. Machine Learning (ML) is used to train the program in identifying number plates. It is done by Anaconda.