A Character Attention Generative Adversarial Network for Degraded Historical Document Restoration.

Document image enhancement plays a crucial role as a pre-processing step in many automated document analysis and recognition tasks such as character recognition. ... Digitized documents such as scientific articles, tax forms, invoices, contract papers, historic texts are widely used nowadays. ... This network is based on conditional GANs, cGANs, network to restore severely degraded document images. ...

arXiv:2112.02719v4 fatcat:sznkn6vkr5fabag2pmaff6bhky

Multiple Versions

Nowadays, deep learning methods are employed in a broad range of research fields. The analysis and recognition of historical documents, as we survey in this work, is not an exception. ... Our study analyzes the papers published in the last few years on this topic from different perspectives: we first provide a pragmatic definition of historical documents from the point of view of the research ... [42] propose a character attention generative adversarial network named CAGAN for restoring heavily degraded character patterns in historical documents so that OCRs can improve their accuracy and help ...

doi:10.3390/jimaging6100110 pmid:34460551 pmcid:PMC8321201 fatcat:nevh2ctshzfwtey4girgjtaftq

DOAJ

This work proposes an end-to-end model for document deblurring using cycle-consistent adversarial networks. ... Our method, named "Blur2Sharp CycleGAN, " generates a sharp image from a blurry one and shows how cycle-consistent generative adversarial networks (CycleGAN) can be used in document deblurring. ... In addition, we gratefully acknowledge the support of NVIDIA Corporation with the donation of the Quadro P6000 GPU used for this research. ...

doi:10.2991/ijcis.d.210407.001 fatcat:7724ts57tvddbaycu5ydr33qqe

DOAJ

In this paper, we propose an effective end-to-end framework named Document Enhancement Generative Adversarial Networks (DE-GAN) that uses the conditional GANs (cGANs) to restore severely degraded document ... To the best of our knowledge, this practice has not been studied within the context of generative adversarial deep networks. ... CONCLUSION In this paper we proposed a Document Enhancement Generative Adversarial Network named DE-GAN to restore severely degraded document images. ...

doi:10.1109/tpami.2020.3022406 pmid:32894707 fatcat:qid525s3wzdfxdobo26y6fr3z4

Multiple Versions

In this paper, we propose an end-to-end trainable framework for restoring historical documents content that follows the correct reading order. ... We then use Hough transform for line detection on the binary mask and combine character results with the layout information to restore document content. ... [9] proposed a character attention generative adversarial network to create high visibility images from severely degraded or low visibility input images. ...

arXiv:2007.06890v1 fatcat:rb3re4ajxvgg5dq7gl7mvbovoi

To tackle these problems, we propose a Feature-Based Fusion Adversarial Recurrent Neural Networks (FARNN-Att) integrated model with an attention mechanism. ... Finally, a regularization method of adversarial training is used to improve the robustness and generalization ability of the model. ... ACKNOWLEDGMENTS The authors would like to thank the providers of the public datasets and the anonymous reviewers for their constructive comments. ...

doi:10.1109/access.2019.2940506 fatcat:ipgm3eqisfdujayrhthjv66doa

DOAJ

Each of the pre-text objectives is specifically tailored for the final downstream tasks. We conduct several ablation experiments that show the importance of each degradation for a specific domain. ... In this work, we propose Text-Degradation Invariant Auto Encoder (Text-DIAE) aimed to solve two tasks, text recognition (handwritten or scene-text) and document image enhancement. ... Other state-of-the-art techniques [37, 69, 70, 80] used conditional-Generative Adversarial Network (c-GAN) based approaches to design a generator which produces the enhanced version of the document while ...

arXiv:2203.04814v3 fatcat:w7vbzopsxngu7njknd6a2n7shi

Open Access Multiple Versions

Paper-intensive industries like insurance, law, and government have long leveraged optical character recognition (OCR) to automatically transcribe hordes of scanned documents into text strings for downstream ... We devise a method to programmatically assemble real text images and real artifacts into realistic-looking "dirty" text images, and use them to train an artifact segmentation network in a weakly supervised ... The results discussed in this letter and references to terms architecture, robustness, efficient, accurate, and bias are with respect to the letters mathematical treatment of a generalized methodology ...

arXiv:1910.07070v3 fatcat:acann2xosndwddx2kqlseuehwa

Multiple Versions

Paper-intensive industries like insurance, law, and government have long leveraged optical character recognition (OCR) to automatically transcribe hordes of scanned documents into text strings for downstream ... We devise a method to programmatically assemble real text images and real artifacts into realistic-looking "dirty" text images, and use them to train an artifact segmentation network in a weakly supervised ... The historical document text characters in [16] are printed while the annotations are handwritten, and the annotations have a slightly different shade, both of which are telltale signs for the network ...

doi:10.1109/wacv45572.2020.9093532 dblp:conf/wacv/QiHLD20 fatcat:zfu5o6s7lreavb2p22jmyiyz4a

It uses a disk-shaped structuring element, whose radius is computed by the minimum entropy-based stroke width transform (SWT). ... AbstractBinarization plays an important role in document analysis and recognition (DAR) systems. ... [42] propose a supervised binarization for historical document images based on hierarchical deep supervised networks (DSNs). ...

doi:10.1186/s13640-021-00556-4 fatcat:aawmkhjf3ngs3bb3zbyyi4xz5q

DOAJ

OCR (Optical Character Recognition) error metrics are also used to gauge the success of the suggested approach. ... Document image warping problem refers to the process of geometrically transforming 2D images. In this work we aim to solve the warped document image problems. ... Generative Adversarial Network (CGAN) and Pix2pix networks [35] fail to produce the document of size of 256x256. ...

doi:10.21608/ijci.2021.53176.1036 fatcat:bcvzyf557zfrnasc2djynymp6y

A visually direct technique for assisting resilience planners is to establish a "resilience profile" for key functions within critical systems. ... This article offers an operational framework that can prove useful to the Department of Homeland Security (DHS) and stakeholders at all levels, both public and private, as a basis for incorporating resilience ... Adversary Attack Path: This provides a graphical depiction of essential steps and stages included in a generic terrorist attack scenario. ...

doi:10.2202/1547-7355.1675 fatcat:j6pgrrlfoff6ln6rim5gzrsjge

At the end of this paper, we conclude the further development of BMs in a more general view. ... With the rapid development of deep learning, training Big Models (BMs) for multiple downstream tasks becomes a popular paradigm. ... Caffe is good for general purpose convolutional neural networks but may not be a good option for recurrent networks due to insufficient relevant documentation. ...

arXiv:2203.14101v4 fatcat:rdikzudoezak5b36cf6hhne5u4

Multiple Versions

Citation

Sha Yuan, Hanyu Zhao, Shuai Zhao, Jiahong Leng, Yangxiao Liang, Xiaozhi Wang, Jifan Yu, Xin Lv, Zhou Shao, Jiaao He, Yankai Lin, Xu Han, Zhenghao Liu, Ning Ding, Yongming Rao, Yizhao Gao, Liang Zhang, Ming Ding, Cong Fang, Yisen Wang, Mingsheng Long, Jing Zhang, Yinpeng Dong, Tianyu Pang, Peng Cui, Lingxiao Huang, Zheng Liang, Huawei Shen, Hui Zhang, Quanshi Zhang, Qingxiu Dong, Zhixing Tan, Mingxuan Wang, Shuo Wang, Long Zhou, Haoran Li, Junwei Bao, Yingwei Pan, Weinan Zhang, Zhou Yu, Rui Yan, Chence Shi, Minghao Xu, Zuobai Zhang, Guoqiang Wang, Xiang Pan, Mengjie Li, Xiaoyu Chu, Zijun Yao, Fangwei Zhu, Shulin Cao, Weicheng Xue, Zixuan Ma, Zhengyan Zhang, Shengding Hu, Yujia Qin, Chaojun Xiao, Zheni Zeng, Ganqu Cui, Weize Chen, Weilin Zhao, Yuan Yao, Peng Li, Wenzhao Zheng, Wenliang Zhao, Ziyi Wang, Borui Zhang, Nanyi Fei, Anwen Hu, Zenan Ling, Haoyang Li, Boxi Cao, Xianpei Han, Weidong Zhan, Baobao Chang, Hao Sun, Jiawen Deng, Chujie Zheng, Juanzi Li, Lei Hou, Xigang Cao, Jidong Zhai, Zhiyuan Liu, Maosong Sun, Jiwen Lu, Zhiwu Lu, Qin Jin, Ruihua Song, Ji-Rong Wen, Zhouchen Lin, Liwei Wang, Hang Su, Jun Zhu, Zhifang Sui, Jiajun Zhang, Yang Liu, Xiaodong He, Minlie Huang, Jian Tang, Jie Tang. "A Roadmap for Big Model." arXiv (2022)

Historically speaking, to a certain limit the nature was capable to restore its functioning to some degree. ... An impact of cyberspace in general and cyberwars in particular on human behavior deserves a special attention. ...

doi:10.28933/ijsr-2019-01-2805 fatcat:fpuhw2ratvaejislgltdowyswi

In the network society power elite does not consist only of governments, but also of non-governmental sector. 16 Furthermore, the US Department of Defence now characterizes information operations as ... , waged between competitors, adversaries or enemies using information means to achieve their objectives". 8 Nevertheless, for M. ...

doi:10.15388/zt/jr.2015.8.8844 fatcat:ky5nabl76fdunksov7zxxitmwu

Szczepanski

A Survey on Deep learning based Document Image Enhancement [article]

Preserved Fulltext

Other Versions

Deep Learning for Historical Document Analysis and Recognition—A Survey

Preserved Fulltext

Blur2Sharp: A GAN-Based Model for Document Image Deblurring

Preserved Fulltext

DE-GAN: A Conditional Generative Adversarial Network for Document Enhancement

Preserved Fulltext

Joint Layout Analysis, Character Detection and Recognition for Historical Document Digitization [article]

Preserved Fulltext

Feature-Based Fusion Adversarial Recurrent Neural Networks for Text Sentiment Classification

Preserved Fulltext

Text-DIAE: Degradation Invariant Autoencoders for Text Recognition and Document Enhancement [article]

Preserved Fulltext

Other Versions

DeepErase: Weakly Supervised Ink Artifact Removal in Document Text Images [article]

Preserved Fulltext

Other Versions

DeepErase: Weakly Supervised Ink Artifact Removal in Document Text Images

Preserved Fulltext

An enhanced binarization framework for degraded historical document images

Preserved Fulltext

Warped Document Image Correction Based on Checkboard Pattern and Geometric Transformation

Preserved Fulltext

An Operational Framework for Resilience

Preserved Fulltext

A Roadmap for Big Model [article]

Preserved Fulltext

Other Versions

Cyberspace of the Fourth Scientific and Technological Revolution

Preserved Fulltext

Collective memory as a resource in Russian information warfare against Latvia

Preserved Fulltext