Truck License Plate Recognition Using Convolution Neural Network

Khoo, Boon Keong (2022) Truck License Plate Recognition Using Convolution Neural Network. Final Year Project (Bachelor), Tunku Abdul Rahman University College.

Text Khoo Boon Keong.pdf Restricted to Registered users only Download (2MB)

Abstract

License plate recognition for truck using Convolution Neural Network (CNN) is studied and developed. In this project, the proposed method with models includes the license plate detection model, license plate types of classifiers and segmentation-free license plate recognition model of connectionist temporal classification (CTC) and bidirectional gated recurrent unit (GRU). For extracting the characters from the license plate on images, the license plate number can be recognized without having to segment the region of interest due to that the segmentation-free recognition model can achieve by using connectionist temporal classification (CTC) approach with bidirectional gated recurrent unit (GRU) with relatively accurate results. The accuracy of each individual model is recorded to evaluate the capability of the model separately for ensuring a final good detection and recognition function in the system after integrating all the models. Hence, the total accuracy of 81.73% can be achieved from the experiments on the test dataset. The amount of dataset available to train is around one thousand images with front and back license plates combined for day and night scene.