Cargando…
CasTabDetectoRS: Cascade Network for Table Detection in Document Images with Recursive Feature Pyramid and Switchable Atrous Convolution
Table detection is a preliminary step in extracting reliable information from tables in scanned document images. We present CasTabDetectoRS, a novel end-to-end trainable table detection framework that operates on Cascade Mask R-CNN, including Recursive Feature Pyramid network and Switchable Atrous C...
Autores principales: | , , , , |
---|---|
Formato: | Online Artículo Texto |
Lenguaje: | English |
Publicado: |
MDPI
2021
|
Materias: | |
Acceso en línea: | https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8540682/ https://www.ncbi.nlm.nih.gov/pubmed/34677300 http://dx.doi.org/10.3390/jimaging7100214 |
_version_ | 1784589046202236928 |
---|---|
author | Hashmi, Khurram Azeem Pagani, Alain Liwicki, Marcus Stricker, Didier Afzal, Muhammad Zeshan |
author_facet | Hashmi, Khurram Azeem Pagani, Alain Liwicki, Marcus Stricker, Didier Afzal, Muhammad Zeshan |
author_sort | Hashmi, Khurram Azeem |
collection | PubMed |
description | Table detection is a preliminary step in extracting reliable information from tables in scanned document images. We present CasTabDetectoRS, a novel end-to-end trainable table detection framework that operates on Cascade Mask R-CNN, including Recursive Feature Pyramid network and Switchable Atrous Convolution in the existing backbone architecture. By utilizing a comparativelyightweight backbone of ResNet-50, this paper demonstrates that superior results are attainable without relying on pre- and post-processing methods, heavier backbone networks (ResNet-101, ResNeXt-152), and memory-intensive deformable convolutions. We evaluate the proposed approach on five different publicly available table detection datasets. Our CasTabDetectoRS outperforms the previous state-of-the-art results on four datasets (ICDAR-19, TableBank, UNLV, and Marmot) and accomplishes comparable results on ICDAR-17 POD. Upon comparing with previous state-of-the-art results, we obtain a significant relative error reduction of [Formula: see text] , [Formula: see text] , [Formula: see text] , and [Formula: see text] on the datasets of ICDAR-19, TableBank, UNLV, and Marmot, respectively. Furthermore, this paper sets a new benchmark by performing exhaustive cross-datasets evaluations to exhibit the generalization capabilities of the proposed method. |
format | Online Article Text |
id | pubmed-8540682 |
institution | National Center for Biotechnology Information |
language | English |
publishDate | 2021 |
publisher | MDPI |
record_format | MEDLINE/PubMed |
spelling | pubmed-85406822021-10-28 CasTabDetectoRS: Cascade Network for Table Detection in Document Images with Recursive Feature Pyramid and Switchable Atrous Convolution Hashmi, Khurram Azeem Pagani, Alain Liwicki, Marcus Stricker, Didier Afzal, Muhammad Zeshan J Imaging Article Table detection is a preliminary step in extracting reliable information from tables in scanned document images. We present CasTabDetectoRS, a novel end-to-end trainable table detection framework that operates on Cascade Mask R-CNN, including Recursive Feature Pyramid network and Switchable Atrous Convolution in the existing backbone architecture. By utilizing a comparativelyightweight backbone of ResNet-50, this paper demonstrates that superior results are attainable without relying on pre- and post-processing methods, heavier backbone networks (ResNet-101, ResNeXt-152), and memory-intensive deformable convolutions. We evaluate the proposed approach on five different publicly available table detection datasets. Our CasTabDetectoRS outperforms the previous state-of-the-art results on four datasets (ICDAR-19, TableBank, UNLV, and Marmot) and accomplishes comparable results on ICDAR-17 POD. Upon comparing with previous state-of-the-art results, we obtain a significant relative error reduction of [Formula: see text] , [Formula: see text] , [Formula: see text] , and [Formula: see text] on the datasets of ICDAR-19, TableBank, UNLV, and Marmot, respectively. Furthermore, this paper sets a new benchmark by performing exhaustive cross-datasets evaluations to exhibit the generalization capabilities of the proposed method. MDPI 2021-10-16 /pmc/articles/PMC8540682/ /pubmed/34677300 http://dx.doi.org/10.3390/jimaging7100214 Text en © 2021 by the authors. https://creativecommons.org/licenses/by/4.0/Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/). |
spellingShingle | Article Hashmi, Khurram Azeem Pagani, Alain Liwicki, Marcus Stricker, Didier Afzal, Muhammad Zeshan CasTabDetectoRS: Cascade Network for Table Detection in Document Images with Recursive Feature Pyramid and Switchable Atrous Convolution |
title | CasTabDetectoRS: Cascade Network for Table Detection in Document Images with Recursive Feature Pyramid and Switchable Atrous Convolution |
title_full | CasTabDetectoRS: Cascade Network for Table Detection in Document Images with Recursive Feature Pyramid and Switchable Atrous Convolution |
title_fullStr | CasTabDetectoRS: Cascade Network for Table Detection in Document Images with Recursive Feature Pyramid and Switchable Atrous Convolution |
title_full_unstemmed | CasTabDetectoRS: Cascade Network for Table Detection in Document Images with Recursive Feature Pyramid and Switchable Atrous Convolution |
title_short | CasTabDetectoRS: Cascade Network for Table Detection in Document Images with Recursive Feature Pyramid and Switchable Atrous Convolution |
title_sort | castabdetectors: cascade network for table detection in document images with recursive feature pyramid and switchable atrous convolution |
topic | Article |
url | https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8540682/ https://www.ncbi.nlm.nih.gov/pubmed/34677300 http://dx.doi.org/10.3390/jimaging7100214 |
work_keys_str_mv | AT hashmikhurramazeem castabdetectorscascadenetworkfortabledetectionindocumentimageswithrecursivefeaturepyramidandswitchableatrousconvolution AT paganialain castabdetectorscascadenetworkfortabledetectionindocumentimageswithrecursivefeaturepyramidandswitchableatrousconvolution AT liwickimarcus castabdetectorscascadenetworkfortabledetectionindocumentimageswithrecursivefeaturepyramidandswitchableatrousconvolution AT strickerdidier castabdetectorscascadenetworkfortabledetectionindocumentimageswithrecursivefeaturepyramidandswitchableatrousconvolution AT afzalmuhammadzeshan castabdetectorscascadenetworkfortabledetectionindocumentimageswithrecursivefeaturepyramidandswitchableatrousconvolution |