Abstract:

Malayalam handwritten script recognition is a challenging task due to the language's complex script and diverse writing styles. Recent advancements in deep learning and prompt engineering offer promising solutions to improve recognition accuracy. This study explores the application of prompt engineering techniques to enhance the performance of Malayalam handwritten script recognition models. By designing effective prompts, we aim to leverage the capabilities of pre-trained language models and improve their ability to recognize handwritten Malayalam text. Our approach involves crafting prompts that capture the nuances of the Malayalam script and utilizing them to fine-tune pre-trained models. We evaluate the performance of our approach on a dataset of Malayalam handwritten text and demonstrate significant improvements in recognition accuracy. Our findings highlight the potential of prompt engineering in improving the performance of handwritten script recognition models for languages like Malayalam.