NUMBER PLATE DETECTION USING DEEP NUERAL NETWORK
Automatic vehicle license plate recognition is an important component of modern intelligent transportation systems (ITS). License plate detection is a challenging task when dealing with open environments and images captured from a certain distance by low-cost cameras. An approach has been proposed for detecting license plates based on a Convolutional neural network. Generally vehicle license plate recognition is divided into several steps including license plate extraction, image region which contains a license plate, character segmentation, and character recognition. The proposed system uses bilateral filtering for smoothening and canny edge detection for localization. After the localization, Vehicles number plate region is extracted. Then the characters are segmented and each character is passed to the CNN. The obtained accuracy is compared with accuracy of the GoogleNet. The system is implemented and stimulated in keras, and its performance is tested on real car images. It is observed that the proposed framework will outperform the comparison of Convolutional Neural Network with GoogLeNet interms of accuracy and loss.