Authors:
J. Angelin Jeba, S. Rubin Bose, R. Regin, M.B. Sudhan, S. Suman Rajest, P. Ramesh Babu
Addresses:
Department of Electronics and Communication Engineering, CEG Campus, Anna University, Chennai, Tamil Nadu, India, jebaangelin@gmail.com. Department of Electronics and Communication Engineering, SRM Institute of Science and Technology, Ramapuram, Chennai, Tamil Nadu, India, rubinbos@srmist.edu.in. Department of Computer Science and Engineering, SRM Institute of Science and Technology, Ramapuram, Chennai, Tamil Nadu, India, reginr@srmist.edu.in, sudhanm@srmist.edu.in. Department of Research and Development (R&D) & International Student Affairs (ISA), Dhaanish Ahmed College of Engineering, Chennai, Tamil Nadu, India, sumanrajest414@gmail.com. Department of Computer Science, College of Engineering and Technology, Wollega University, Nekemte, Oromia Region, Ethiopia, rameshbabup@wollegauniversity.edu.et.
Tamil character recognition (TCR) presents a significant challenge due to the intricate shapes and diverse sizes of Tamil characters. This paper proposes a novel TCR method leveraging the state-of-the-art object detection model YOLOv8. By harnessing YOLOv8’s object detection capabilities, our approach accurately identifies Tamil characters within images, ensuring precise recognition of individual characters across varied writing styles and typefaces. We further enhance the recognition accuracy of the CNN model through specific pre-processing steps and transfer learning from pre-trained models. Experimental results on a benchmark dataset of Tamil character images demonstrate the effectiveness of our method, achieving a remarkable precision of 98.74%, recall of 96.63%, and F1 score of 95.52%. Our proposed method not only addresses the challenges associated with TCR but also paves the way for advancements in character recognition techniques, offering a robust solution for real-world applications such as automated text recognition from Tamil documents and signage. This research contributes to the broader field of computer vision and deep learning, providing valuable insights into enhancing the accessibility of Tamil content for native and non-native speakers.
Keywords: Tamil Character Recognition; Yolo-v8 and Deep Learning; Natural Language Processing; Convolutional Neural Network (CNN); Artificial Neural Network (ANN); Handwritten Tamil Character Recognition (HTCR); Data Spilt; Handwritten Character Recognition (HCR).
Received on: 15/08/2023, Revised on: 07/10/2023, Accepted on: 12/11/2023, Published on: 05/03/2024
AVE Trends in Intelligent Computing Systems, 2024 Vol. 1 No. 1, Pages: 1-16