skip to main content
Language:
Search Limited to: Search Limited to: Resource type Show Results with: Show Results with: Search type Index

CurT: End-to-End Text Line Detection in Historical Documents with Transformers

Frontiers in Handwriting Recognition, 2022, Vol.13639, p.34-48 [Peer Reviewed Journal]

The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 ;Distributed under a Creative Commons Attribution 4.0 International License ;ISSN: 0302-9743 ;ISBN: 3031216474 ;ISBN: 9783031216473 ;EISSN: 1611-3349 ;EISBN: 9783031216480 ;EISBN: 3031216482 ;DOI: 10.1007/978-3-031-21648-0_3

Digital Resources/Online E-Resources

Citations Cited by
  • Title:
    CurT: End-to-End Text Line Detection in Historical Documents with Transformers
  • Author: Kiessling, Benjamin
  • Subjects: Computer Science ; Computer Vision and Pattern Recognition ; Document analysis ; Humanities and Social Sciences ; Machine learning ; Object detection ; Text line detection
  • Is Part Of: Frontiers in Handwriting Recognition, 2022, Vol.13639, p.34-48
  • Description: We present the curve transformer (CurT), a novel method of direct baseline detection that models document text line detection as set prediction of cubic Bézier curves, simplifying the layout analysis pipeline by removing the need for the laboriously hand-crafted postprocessing algorithms that are necessary with the current state of the art. CurT combines multiple appealing features: direct prediction enabling processing of material that is ill-suited for the prevailing methods adapting semantic segmentation backbones, a conceptually simple Transformer-based encoder-decoder architecture that can be extended to additional tasks beyond baseline detection, and increased computational efficiency in comparison to older approaches. In addition, we demonstrate that CurT achieves metrics that are competitive with methods based on semantic segmentation. Training and inference code is available under Apache 2.0 license at https://github.com/mittagessen/curt.
  • Publisher: Cham: Springer International Publishing
  • Language: English
  • Identifier: ISSN: 0302-9743
    ISBN: 3031216474
    ISBN: 9783031216473
    EISSN: 1611-3349
    EISBN: 9783031216480
    EISBN: 3031216482
    DOI: 10.1007/978-3-031-21648-0_3
  • Source: Hyper Article en Ligne (HAL) (Open Access)
Back to results list
INSPIRE LIBRARY - TON DUC THANG UNIVERSITY (84-028) 37 755 057 Feedback
19 Nguyen Huu Tho St. Dist.7, HCM thuvien@tdtu.edu.vn

Copyright © 2017 INSPIRE LIBRARY - TON DUC THANG UNIVERSITY

Searching Remote Databases, Please Wait