Fine-Tuned Convolutional Neural Networks for Tamil Handwritten Text Recognition
C. R. Dhivyaa, K. Nithya, R. Dharshini, R. Sudhakar, K. Sathis Kumar, T. Janani · 2023
Tamil Handwritten Text Recognition is the process of identifying and transcribing handwritten Tamil characters into machine-encoded text. Due to variations in the style, stroke, and shape of characters, identifying handwritten text is a critical task. The recognition of Tamil handwritten characters is an essential step in many applications, such as digitizing historical Tamil documents, handwriting recognition in education, and Tamil language translation systems. The development of accurate and efficient recognition systems for Tamil handwritten characters can greatly benefit these applications and contribute to the preservation and dissemination of Tamil literature and culture. The fine-tuned Convolutional Neural Network (CNN) is proposed for the identification of Tamil handwritten text. The proposed approach involves the use of pre-trained CNN models and fine-tuning them on a dataset of Tamil handwritten characters. The dataset consists of a large number of images of individual Tamil characters, including various styles of handwriting. To assess the efficacy of the suggested method, various experiments are performed on optimizing the CNN model's hyper parameters, utilizing both grid search and random search approaches. The results proved that the grid search with CNN model yields better result to recognize the Tamil text from the images.