HANDWRITTEN TEXT RECOGNITION SYSTEM USING DEEP LEARNING

Handwritten text recognition (HTR) remains a significant challenge in the field of computer vision and patternrecognition due to the diverse styles, irregular spacing, and cursive writing in human handwriting. This paperpresents a deep learning-based approach for recognizing handwritten line text using a hybrid architecture thatcombines Convolutional Recurrent Neural Networks (CRNN) and Bidirectional Long Short-Term Memory(BiLSTM) layers. The proposed model leverages convolutional layers to extract spatial features from inputimages, followed by BiLSTM layers that capture temporal dependencies in both forward and backward directions.The final transcription is performed using a Connectionist Temporal Classification (CTC) loss function, allowingalignment-free sequence decoding. The system is trained and evaluated on the IAM dataset, including extendedsupport for cursive writing samples. Experimental results demonstrate promising accuracy and generalizationcapabilities across varying handwriting styles. The model achieves an efficient balance between recognitionperformance and computational complexity, making it suitable for real-world applications such as digital archivalof handwritten documents and automated form processing.