Machine Learning-Based Tamil Handwritten Word Recognition

Recognition of handwritten Tamil words is a difficult problem because of the script’s complexity and the wide variety of handwriting styles. To enhance the recognition of handwritten Tamil characters, we present a pre-processing method based on image segmentation in this work. The proposed technique includes normalization, binarization, noise removal, and image segmentation to identify individual characters. We also employ a recognition system that utilizes a CNN for character and word classification and recognition. We estimate the performance of the recognition system on a dataset of handwritten Tamil characters. Experimental results show that the proposed pre-processing technique significantly improves recognition performance for both handwritten Tamil characters with an 95% of accuracy. Our findings suggest that pre-processing techniques, including image segmentation, can play a critical role in improving the recognition performance of handwritten Tamil characters, when combined with modern deep-learning techniques such as CNN This work provides insights on improving recognition performance for Tamil script and may have the potential to enhance the practical applications of handwriting recognition systems in various domains.