Abstract:The Handwritten Text Recognition problem has been a challenge for researchers for the last few decades, especially in the domain of computer vision, a subdomain of pattern recognition. Variability of texts amongst writers, cursiveness, and different font styles of handwritten texts with degradation of historical text images make it a challenging problem. Recognizing scanned document images in neural network-based systems typically involves a two-step approach: segmentation and recognition. However, this method has several drawbacks. These shortcomings encompass challenges in identifying text regions, analyzing layout diversity within pages, and establishing accurate ground truth segmentation. Consequently, these processes are prone to errors, leading to bottlenecks in achieving high recognition accuracies. Thus, in this study, we present an end-to-end paragraph recognition system that incorporates internal line segmentation and gated convolutional layers based encoder. The gating is a mechanism that controls the flow of information and allows to adaptively selection of the more relevant features in handwritten text recognition models. The attention module plays an important role in performing internal line segmentation, allowing the page to be processed line-by-line. During the decoding step, we have integrated a connectionist temporal classification-based word beam search decoder as a post-processing step. In this work, we have extended existing LexiconNet by carefully applying and utilizing gated convolutional layers in the existing deep neural network. Our results at line and page levels also favour our new GatedLexiconNet. This study reported character error rates of 2.27% on IAM, 0.9% on RIMES, and 2.13% on READ-16, and word error rates of 5.73% on IAM, 2.76% on RIMES, and 6.52% on READ-2016 datasets.

Adversarial Feature Enhancing Network for End-to-End Handwritten Paragraph Recognition

Towards End-to-end Handwritten Document Recognition

A Comprehensive Handwritten Paragraph Text Recognition System: LexiconNet

A Stroke-Based RNN for Writer-Independent Online Signature Verification.

GatedLexiconNet: A Comprehensive End-to-End Handwritten Paragraph Text Recognition System

A New Hybrid-Parameter Recurrent Neural Network for Online Handwritten Chinese Character Recognition

Refocus attention span networks for handwriting line recognition

Deep Network with Pixel-Level Rectification and Robust Training for Handwriting Recognition

Sentence-level Online Handwritten Chinese Character Recognition

From Textline to Paragraph: A Promising Practice for Chinese Text Recognition

An Efficient End-to-End Neural Model for Handwritten Text Recognition

Writer Adaptive Feature Extraction Based on Convolutional Neural Networks for Online Handwritten Chinese Character Recognition

Multi-layer Recurrent Neural Network Based Offline Arabic Handwriting Recognition

Fast writer adaptation with style extractor network for handwritten text recognition

Uncovering the Handwritten Text in the Margins: End-to-end Handwritten Text Detection and Recognition

Robust End-to-End Offline Chinese Handwriting Text Page Spotter with Text Kernel

End to End Recognition System for Recognizing Offline Unconstrained Vietnamese Handwriting

Writer Adaptation Via Deeply Learned Features for Online Chinese Handwriting Recognition

Residual Recurrent Neural Network with Sparse Training for Offline Arabic Handwriting Recognition

Handwritten Mathematical Expression Recognition via Paired Adversarial Learning

Combination of Global and Local Baseline-Independent Features for Offline Arabic Handwriting Recognition.