I did my undergraduate thesis on Bangla License Plate Recognition using Convolutional Neural Network(CNN). It took me almost fifteen/sixteen months to complete this project. Since there was no data-set available for the Bangla license plate. So, I had to spend almost three months to make the data-set. I captured nearly three thousand photographs of the car, auto-rickshaw, truck, and motorcycle and processed them manually for creating a standard data set. Moreover, I had no teammates to help me. So, it was very challenging for me to prepare a data-set, setup work environment, training, and running the test and writing report.

Here is the abstract of my work:
In the last few years, the deep learning technique in particular Convolutional Neural Networks (CNNs) are using massively in the field of computer vision and machine learning. This deep learning technique provides state-of-the-art accuracy in different classification tasks on different benchmarks such as MNIST, CIFAR-10, CIFAR-100, ImageNet. However, there is a lot of research has been conducted for Bangla License plate recognition in the last decade. None of them are used to deploy a physical system for the Bangla license plate recognition system because of their poor recognition accuracy. In this thesis, we would like to implement CNNs based Bangla license plate recognition system with better accuracy that can be used for different purposes like roadside assistance, automatic parking lot management system and so on

Full thesis work was published on 24 Dec 2017 at BRAC University digital thesis archive. You can access it from here
A brief version of this work was published on 4th Sep 2018 by arXiv.org. Now it’s available here.
ResearchGate Link: Bangla License Plate Recognition Using Convolutional Neural Networks (CNN)

I had to present my thesis in the panel of Mahbub Majumdar, Ph.D. M. Mojumdar spent his education life at the University of Cambridge (Ph.D. Mathematics), Stanford University (M.S.), Massachusetts Institute of Technology (S.B.). So, it was really challenging for me to present my work. The question-answer session after the presentation was longer than the presentation time. I got a full grade in my thesis.

I am grateful to my thesis supervisor  Moin Mostakim. I am also thankful to my sister Shamima Nasrin, my brother-in-law Zahangir Alom for helping me to complete the whole thing within a limited time frame.