Handwritten Text Recognition Using Deep Learning:A Comprehensive Review

作     者：Khan, Mohd Ayan Arif, Hamza Singh, Vishal Sharma, Dolly 

作者机构：Department of Computer Science and Engineering Amity School of Engineering and Technology Amity University Uttar Pradesh Noida201313 India 

出 版 物：《SSRN》 

年 卷 期：2023年

核心收录：

主　　题：Convolutional neural networks 

摘      要：Deep learning techniques and methods have shown excellent development in the field of offline/online handwritten text recognition(OHTR) in recent times. There are many reviews in the realm of OHTR, but none of them provides a collective and comprehensivereview of these many diverse technologies and methods. This review paper provides a profound and comprehensive scientific reviewof various techniques, models, algorithms, and architectures in the field of OHTR while highlighting the advancements that havebeen made in this area of deep learning. In this review, native Convolutional Neural Networks (CNN), hybrid models, and machinelearning classifiers have been noticed to have high potential and have shown high recognition accuracy in many languages andscripts. This review is divided into four sections: Traditional and Hybrid models;New Approaches based on Neural Networks, NewApproaches based on Neural Networks;Image Enhancement, Feature Extraction and Noise Reduction;Innovative Approaches andNovel Techniques where each category comprises of a set of advancement with a respective advancement and problems in *** also lists the advantages and applications of some of the models. This review certainly gives the overall idea of various optionsavailable for OHTR, and it helps to understand the fundamental difference between different approaches and can help to choose thebest if one is looking to use the concept of transfer learning. © 2023, The Authors. All rights reserved.