Chinese character recognition system using Convolutional Neural Network (CNN) / Nur Fatihah Hatini Mat Sam
This study aims to develop a Chinese character recognition system using Convolutional Neural Network (CNN) system that is specifically designed for the challenging task of recognizing Chinese characters, taking into consideration the complex structures and various styles of writing that these charac...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
2024
|
| Subjects: | |
| Online Access: | https://ir.uitm.edu.my/id/eprint/95736/1/95736.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | This study aims to develop a Chinese character recognition system using Convolutional Neural Network (CNN) system that is specifically designed for the challenging task of recognizing Chinese characters, taking into consideration the complex structures and various styles of writing that these characters possess. With the Chinese language's global importance, particularly in places like Malaysia with one of the large Chinese communities, there is a strong need for an effective and user-friendly character recognition system. The CNN system is remarkably accurate, with a recognition rate of 99.99% for Chinese character recognition and a training accuracy of 92%. The research includes an in-depth examination of CNN requirements, the methodical building of a recognition system, and an accurate evaluation of its performance using testing measures such as precision, recall, and F1-score. The results indicate impressive testing accuracy with minimal loss, demonstrating the system's ability to generalize to newly collected data. The confusion matrix provides an in-depth analysis of the model's classification accuracy over 135 different Chinese character classes. Functionality testing verifies the system's ability to accommodate different handwriting styles, increasing its value for learners of varying competence levels. Recognizing limitations due to the intentional focus on recognizable characters for beginners, the study proposes future directions, such as investigating advanced CNN architectures, augmenting training datasets, incorporating transfer learning strategies, and incorporating user feedback for continuous refinement. In short, this initiative aims to bridge the gap between technological breakthroughs in CNNs and evolving educational objectives for Chinese character recognition, therefore contributing to the larger conversation about the connection of technology and education in linguistic domains. |
|---|