Handwritten Text Recognition Using CNN-RNN Hybrid Model

Naman Shah; Lalit Purohit; Mukesh Sakle

doi:10.36948/ijfmr.2025.v07i06.60351

Handwritten Text Recognition Using CNN-RNN Hybrid Model

Author(s)	Mr. Naman Shah, Prof. Lalit Purohit, Prof. Mukesh Sakle
Country	India
Abstract	Handwritten Text Recognition (HTR) has long stood as a challenging domain in computer vision and pattern recognition, particularly due to the variability in individual handwriting styles, noise in scanned documents, and differences in word structure. In this study, we present a CNN-RNN hybrid model enhanced with a Connectionist Temporal Classification (CTC) loss function, aimed at improving the accuracy and reliability of offline handwritten text recognition. By leveraging convolutional layers for spatial feature extraction and bidirectional LSTM layers for sequential modeling, our approach balances spatial and temporal understanding effectively. We trained our model using the IAM dataset, incorporating careful preprocessing and character filtering techniques. Experimental results demonstrate that our model achieves an accuracy of 87.5% on a 200-sample validation set, a notable improvement over our baseline of 43.33%. Compared to traditional techniques and several recent deep learning-based models, our pipeline maintains a strong performance while retaining architectural simplicity. We also analyze word-level precision, recall, and F1-score, and present a detailed comparison against recent advancements in the field. This work contributes a reliable and efficient HTR pipeline with room for improvement via ensemble learning and language model integration.
Keywords	Handwritten Text Recognition, CNN-RNN Hybrid, CTC Loss, IAM Dataset, Beam Search Decoding, Deep Learning, Sequence Modeling, Optical Character Recognition, LSTM
Field	Engineering
Published In	Volume 7, Issue 6, November-December 2025
Published On	2025-12-05
DOI	https://doi.org/10.36948/ijfmr.2025.v07i06.60351

View / Download PDF File

E-ISSN 2582-2160

doi

CrossRef DOI is assigned to each research paper published in our journal.

IJFMR DOI prefix is
10.36948/ijfmr

Downloads

Research Paper Format Copyright Permission Form and Undertaking Form Cover Page Vol 8 Isu 2 Cover Page Vol 8 Isu 1 Cover Page Vol 7 Isu 6

All research papers published on this website are licensed under Creative Commons Attribution-ShareAlike 4.0 International License, and all rights belong to their respective authors/researchers.

CC-BY-SA

About IJFMR Fees & Payment Current Issue Publication Archive	Submit Research Paper Track Submission Status Publication Guidelines Publication Ethics Peer Review & Plagiarism	Join as a Reviewer Editors & Reviewers Reviewer Referral Program Get Reviewer Membership Certi.	Website/Journal Policies Usage Policy Content Policies Privacy Policy

Contact Us		+91-9687-828-838	editor@ijfmr.com

International Journal For Multidisciplinary Research

A Widely Indexed Open Access Peer Reviewed Multidisciplinary Bi-monthly Scholarly International Journal

Handwritten Text Recognition Using CNN-RNN Hybrid Model

Share this