Vehicle Identification using coordination of CCTV Cameras

Abstract

Vehicle identification using coordination of CCTV cameras are used for Identification of vehicle, measure the speed of a vehicle and to find exactly which route does a vehicle used. This goals are proved both theoretical and practical way. SUMO simulator used for theoretical approach to estimate the vehicle location inside Nirma University. Estimation of the vehicle location using vehicle density and OSM is used for Identification of vehicle, measuring the speed of the vehicle and to find exactly which route does the vehicle use. For computing the optimal distance, a database has been created which includes all the routes of Nirma University. Nirma University road network has been constructed using OSM. SUMO simulator helps in the simulation process. SUMO simulator imports the .osm file which includes the road network in which the vehicles are running. Different dense scenarios are taken for the simulation. Simulation gives the vehicle parameters like vehicle id, speed, position, location etc. information. So using these vehicle parameters one can identify the vehicle. For practical approach, Mask RCNN has been used to detect and classify the vehicles. Mask R-CNN generate the three outputs for each candidate object. First is a class label, second is a bounding box offset and third is object mask. For object recognition COCO API is used. Speed of the vehicles have been also detected.