Enhancing Handwritten Character Recognition with XGBoost: A Machine Learning Approach

Recognizing handwritten text poses a significant challenge due to the distinct characteristics inherent in each individual’s handwriting. Consequently, recognition systems must exhibit adaptability to identify similar or dissimilar characters with varying traits. Particularly challenging is the recognition of characters in ancient documents, where the presence of numerous types of noise introduces complexity. When digitizing these historical records, diverse noise patterns emerge, significantly impacting recognition systems. Hence, the essential elements of a successful recognition system encompass the conversion of historical documents into digital format, the application of suitable preprocessing methods, and the utilisation of a resilient classifier. This study presents a thorough system for recognising handwritten characters in English and Telugu, which utilises a deep convolutional neural network. The proposed system achieves an impressive accuracy of 95.20% with minimal model loss, underscoring its efficiency and effectiveness in handling the complexities of handwritten character recognition, especially in the context of ancient documents.