dhSegment: A Generic Deep-Learning Approach for Document Segmentation. Ares Oliveira, S., Seguin, B., & Kaplan, F. 2018 16th International Conference on Frontiers in Handwriting Recognition (ICFHR), August, 2018. Conference Name: 2018 16th International Conference on Frontiers in Handwriting Recognition (ICFHR) ISBN: 9781538658758 Place: Niagara Falls, NY, USA Publisher: IEEE![dhSegment: A Generic Deep-Learning Approach for Document Segmentation [link]](https://bibbase.org/img/filetypes/link.svg "link")
Paper doi abstract bibtex In recent years there have been multiple successful attempts tackling document processing problems separately by designing task specific hand-tuned strategies. We argue that the diversity of historical document processing tasks prohibits to solve them one at a time and shows a need for designing generic approaches in order to handle the variability of historical series. In this paper, we address multiple tasks simultaneously such as page extraction, baseline extraction, layout analysis or multiple typologies of illustrations and photograph extraction. We propose an open-source implementation of a CNN-based pixel-wise predictor coupled with task dependent post-processing blocks. We show that a single CNN-architecture can be used across tasks with competitive results. Moreover most of the task-specific post-precessing steps can be decomposed in a small number of simple and standard reusable operations, adding to the flexibility of our approach.
@article{ares_oliveira_dhsegment_2018,
title = {{dhSegment}: {A} {Generic} {Deep}-{Learning} {Approach} for {Document} {Segmentation}},
shorttitle = {{dhSegment}},
url = {https://ieeexplore.ieee.org/document/8563218/},
doi = {10.1109/ICFHR-2018.2018.00011},
abstract = {In recent years there have been multiple successful attempts tackling document processing problems separately by designing task specific hand-tuned strategies. We argue that the diversity of historical document processing tasks prohibits to solve them one at a time and shows a need for designing generic approaches in order to handle the variability of historical series. In this paper, we address multiple tasks simultaneously such as page extraction, baseline extraction, layout analysis or multiple typologies of illustrations and photograph extraction. We propose an open-source implementation of a CNN-based pixel-wise predictor coupled with task dependent post-processing blocks. We show that a single CNN-architecture can be used across tasks with competitive results. Moreover most of the task-specific post-precessing steps can be decomposed in a small number of simple and standard reusable operations, adding to the flexibility of our approach.},
urldate = {2023-05-03},
journal = {2018 16th International Conference on Frontiers in Handwriting Recognition (ICFHR)},
author = {Ares Oliveira, Sofia and Seguin, Benoit and Kaplan, Frederic},
month = aug,
year = {2018},
note = {Conference Name: 2018 16th International Conference on Frontiers in Handwriting Recognition (ICFHR)
ISBN: 9781538658758
Place: Niagara Falls, NY, USA
Publisher: IEEE},
pages = {7--12},
}
Downloads: 0
{"_id":"z5xH6yZGSGMoYL2Gg","bibbaseid":"aresoliveira-seguin-kaplan-dhsegmentagenericdeeplearningapproachfordocumentsegmentation-2018","author_short":["Ares Oliveira, S.","Seguin, B.","Kaplan, F."],"bibdata":{"bibtype":"article","type":"article","title":"dhSegment: A Generic Deep-Learning Approach for Document Segmentation","shorttitle":"dhSegment","url":"https://ieeexplore.ieee.org/document/8563218/","doi":"10.1109/ICFHR-2018.2018.00011","abstract":"In recent years there have been multiple successful attempts tackling document processing problems separately by designing task specific hand-tuned strategies. We argue that the diversity of historical document processing tasks prohibits to solve them one at a time and shows a need for designing generic approaches in order to handle the variability of historical series. In this paper, we address multiple tasks simultaneously such as page extraction, baseline extraction, layout analysis or multiple typologies of illustrations and photograph extraction. We propose an open-source implementation of a CNN-based pixel-wise predictor coupled with task dependent post-processing blocks. We show that a single CNN-architecture can be used across tasks with competitive results. Moreover most of the task-specific post-precessing steps can be decomposed in a small number of simple and standard reusable operations, adding to the flexibility of our approach.","urldate":"2023-05-03","journal":"2018 16th International Conference on Frontiers in Handwriting Recognition (ICFHR)","author":[{"propositions":[],"lastnames":["Ares","Oliveira"],"firstnames":["Sofia"],"suffixes":[]},{"propositions":[],"lastnames":["Seguin"],"firstnames":["Benoit"],"suffixes":[]},{"propositions":[],"lastnames":["Kaplan"],"firstnames":["Frederic"],"suffixes":[]}],"month":"August","year":"2018","note":"Conference Name: 2018 16th International Conference on Frontiers in Handwriting Recognition (ICFHR) ISBN: 9781538658758 Place: Niagara Falls, NY, USA Publisher: IEEE","pages":"7–12","bibtex":"@article{ares_oliveira_dhsegment_2018,\n\ttitle = {{dhSegment}: {A} {Generic} {Deep}-{Learning} {Approach} for {Document} {Segmentation}},\n\tshorttitle = {{dhSegment}},\n\turl = {https://ieeexplore.ieee.org/document/8563218/},\n\tdoi = {10.1109/ICFHR-2018.2018.00011},\n\tabstract = {In recent years there have been multiple successful attempts tackling document processing problems separately by designing task specific hand-tuned strategies. We argue that the diversity of historical document processing tasks prohibits to solve them one at a time and shows a need for designing generic approaches in order to handle the variability of historical series. In this paper, we address multiple tasks simultaneously such as page extraction, baseline extraction, layout analysis or multiple typologies of illustrations and photograph extraction. We propose an open-source implementation of a CNN-based pixel-wise predictor coupled with task dependent post-processing blocks. We show that a single CNN-architecture can be used across tasks with competitive results. Moreover most of the task-specific post-precessing steps can be decomposed in a small number of simple and standard reusable operations, adding to the flexibility of our approach.},\n\turldate = {2023-05-03},\n\tjournal = {2018 16th International Conference on Frontiers in Handwriting Recognition (ICFHR)},\n\tauthor = {Ares Oliveira, Sofia and Seguin, Benoit and Kaplan, Frederic},\n\tmonth = aug,\n\tyear = {2018},\n\tnote = {Conference Name: 2018 16th International Conference on Frontiers in Handwriting Recognition (ICFHR)\nISBN: 9781538658758\nPlace: Niagara Falls, NY, USA\nPublisher: IEEE},\n\tpages = {7--12},\n}\n\n","author_short":["Ares Oliveira, S.","Seguin, B.","Kaplan, F."],"key":"ares_oliveira_dhsegment_2018","id":"ares_oliveira_dhsegment_2018","bibbaseid":"aresoliveira-seguin-kaplan-dhsegmentagenericdeeplearningapproachfordocumentsegmentation-2018","role":"author","urls":{"Paper":"https://ieeexplore.ieee.org/document/8563218/"},"metadata":{"authorlinks":{}},"html":""},"bibtype":"article","biburl":"https://bibbase.org/zotero/fsimonetta","dataSources":["pzyFFGWvxG2bs63zP"],"keywords":[],"search_terms":["dhsegment","generic","deep","learning","approach","document","segmentation","ares oliveira","seguin","kaplan"],"title":"dhSegment: A Generic Deep-Learning Approach for Document Segmentation","year":2018}