Special Character Recognition Using Deep Learning

Date
2022
Authors
Li, Changjian
Supervisor
Yan, Wei Qi
Item type
Thesis
Degree name
Master of Computer and Information Sciences
Journal Title
Journal ISSN
Volume Title
Publisher
Auckland University of Technology
Abstract

In recent years, deep learning methods have been applied to our daily lives and various industries. Visual object detection methods are broadly employed to a consortium of tasks, including human face detection in public areas, traffic signs detection, car plate number recognition, etc. Natural Language Processing (NLP) methods are implemented for language translation, Automatic Speech Recognition (ASR), client embedding, item embedding, etc. In this thesis, we contribute to special character recognition by using deep learning. The Adaptive Bezier Curve Network (ABCNet) is a text detection and recognition method utilized to recognize English Braille, which implements parameterized Bezier curves for detecting arbitrary-shape text in natural scenes. YOLOv5 is the second deep learning method that was implemented for Māori symbol recognition. The methods show outstanding performance in our experiments. Both methods detect and recognize visual objects with high accuracies. The results of our experiments prove deep learning methods are feasible to be implemented for detecting and classifying special characters, shortening the time cost of translation, and reducing labor costs.

Description
Keywords
Deep learning , Object detection , Scene text recognition and detection , ABCNet , YOLOv5
Source
DOI
Publisher's version
Rights statement
Collections