This paper introduces an advanced end-to-end framework for Handwritten Text Recognition (HTR) that effectively bridges the gap between traditional optical character recognition and modern deep learning. The system utilizes a novel fusion of Bidirectional Long Short-Term Memory (BiLSTM) networks and Convolutional Neural Networks (CNNs) to capture both complex spatial features and sequential dependencies in unsegmented handwritten scripts. By implementing a script-independent neural architecture, the model achieves state-of-the-art accuracy across diverse datasets, significantly reducing error rates in the digitization of historical and legal documents. This research provides a scalable and robust solution for real-time transcription, paving the way for improved accessibility and automated archival systems in the digital era.
2025 Conference Paper, IEEE.
https://www.researchgate.net/publication/394686471_Advanced_Handwritten_Text_Recognition_and_Analysis_System