Vision Based Vehicle Information Inspection System Using Deep Learning

The proposed Vehicle Information Inspection System (VIIS) includes inspection of the vehicle color, type, and license plate. The aim of this paper is to recognize the moving vehicle's color, type, and license plate characters. The vehicle color is recognized using K-means clustering method and K-Nearest Neighbors (K-NN) method. The vehicle type is recognized using Convolutional Neural Network (CNN). The proposed vehicle license plate recognition system consists of four processes: license plate localization, license plate skew correction, character segmentation, and character recognition. CNN is also mainly used to recognize license plate characters. The extracted vehicle information is stored in the database. This information is matched and inspected with the blacklist vehicle information in the database. The proposed method can help in monitoring and inspection of the blacklist vehicles on the road without any human effort.