Handwritten Text Recognition using Hybrid CNN-GRU Model and CNN-LSTM Model on Parzival Database: A Novel Approach
Madhav Sharma · Advances in engineering research/Advances in Engineering Research · 2024
The use of deep learning models, especially Convolutional Neural Networks and Gated Recurrent Units , has been a popular approach to improve the performance of handwriting recognition (HTR) algorithms In this research, perzival dataset is used it evaluates the performance of the HTR system incorporating CNN and GRU as a reference.Our results show that this integration significantly reduces loss and improves the HTR process, demonstrating the effectiveness of deep learning models used for HTR implementation.This study proposes a new method for handwritten text recognition (HTR) using a hybrid of CNN-GRU and CNN-LSTM models on the Parzival database.CNN-GRU and CNN-LSTM models were used to extract spatial and temporal features from the input images, respectively.The analysis showed that the CNN-GRU model suffered less loss compared to the CNN-LSTM model, indicating better performance.The proposed method provides a promising strategy to improve the accuracy of HTR hybrid modeling, and has the potential to be applied to various applications such as document digitization.