ACMU-nets: Attention cascading modular U-nets incorporating squeeze and excitation blocks

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

In document analysis research, image-to-image conversion models such as a U-Net have been shown significant performance. Recently, cascaded U-Nets research is suggested for solving complex document analysis studies. However, improving performance by adding U-Net modules requires using too many parameters in cascaded U-Nets. Therefore, in this paper, we propose a method for enhancing the performance of cascaded U-Nets. We suggest a novel document image binarization method by utilizing Cascading Modular U-Nets (CMU-Nets) and Squeeze and Excitation blocks (SE-blocks). Through verification experiments, we point out the problems caused by the use of SE-blocks in existing CMU-Nets and suggest how to use SE-blocks in CMU-Nets. We use the Document Image Binarization (DIBCO) 2017 dataset to evaluate the proposed model.

Original languageEnglish
Title of host publicationDocument Analysis Systems - 14th IAPR International Workshop, DAS 2020, Proceedings
EditorsXiang Bai, Dimosthenis Karatzas, Daniel Lopresti
PublisherSpringer
Pages118-130
Number of pages13
ISBN (Print)9783030570576
DOIs
Publication statusPublished - 2020
Event14th IAPR International Workshop on Document Analysis Systems, DAS 2020 - Wuhan, China
Duration: Jul 26 2020Jul 29 2020

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume12116 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference14th IAPR International Workshop on Document Analysis Systems, DAS 2020
CountryChina
CityWuhan
Period7/26/207/29/20

All Science Journal Classification (ASJC) codes

  • Theoretical Computer Science
  • Computer Science(all)

Fingerprint Dive into the research topics of 'ACMU-nets: Attention cascading modular U-nets incorporating squeeze and excitation blocks'. Together they form a unique fingerprint.

Cite this