Unsupervised Adaptation for Synthetic-to-Real Handwritten Word Recognition

被引:0
|
作者
Kang, Lei [1 ,2 ]
Rusinol, Marcal [1 ]
Fornes, Alicia [1 ]
Riba, Pau [1 ]
Villegas, Mauricio [2 ]
机构
[1] Univ Autonoma Barcelona, Comp Vis Ctr, Barcelona, Spain
[2] Omni Us, Berlin, Germany
关键词
WRITER ADAPTATION;
D O I
暂无
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Handwritten Text Recognition (HTR) is still a challenging problem because it must deal with two important difficulties: the variability among writing styles, and the scarcity of labelled data. To alleviate such problems, synthetic data generation and data augmentation are typically used to train HTR systems. However, training with such data produces encouraging but still inaccurate transcriptions in real words. In this paper, we propose an unsupervised writer adaptation approach that is able to automatically adjust a generic handwritten word recognizer, fully trained with synthetic fonts, towards a new incoming writer. We have experimentally validated our proposal using five different datasets, covering several challenges (i) the document source: modern and historic samples, which may involve paper degradation problems; (ii) different handwriting styles: single and multiple writer collections; and (iii) language, which involves different character combinations. Across these challenging collections, we show that our system is able to maintain its performance, thus, it provides a practical and generic approach to deal with new document collections without requiring any expensive and tedious manual annotation step.
引用
收藏
页码:3491 / 3500
页数:10
相关论文
共 50 条
  • [21] Multimodal Generation of Novel Action Appearances for Synthetic-to-Real Recognition of Activities of Daily Living
    Marinov, Zdravko
    Schneider, David
    Roitberg, Alina
    Stiefelhagen, Rainer
    2022 IEEE/RSJ INTERNATIONAL CONFERENCE ON INTELLIGENT ROBOTS AND SYSTEMS (IROS), 2022, : 11320 - 11327
  • [22] Robust Synthetic-to-Real Transfer for Stereo Matching
    Zhang, Jiawei
    Li, Jiahe
    Huang, Lei
    Yu, Xiaohan
    Gu, Lin
    Zheng, Jin
    Bai, Xiao
    2024 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR), 2024, : 20247 - 20257
  • [23] Synthetic-to-Real Pose Estimation with Geometric Reconstruction
    Lin, Qiuxia
    Gu, Kerui
    Yang, Linlin
    Yao, Angela
    ADVANCES IN NEURAL INFORMATION PROCESSING SYSTEMS 36 (NEURIPS 2023), 2023,
  • [24] Handwritten word preprocessing for database adaptation
    Oprean, Cristina
    Likforman-Sulem, Laurence
    Mokbel, Chafic
    DOCUMENT RECOGNITION AND RETRIEVAL XX, 2013, 8658
  • [25] A recognition and verification strategy, for handwritten word recognition
    Morita, M
    Sabourin, R
    Bortolozzi, F
    Suen, CY
    SEVENTH INTERNATIONAL CONFERENCE ON DOCUMENT ANALYSIS AND RECOGNITION, VOLS I AND II, PROCEEDINGS, 2003, : 482 - 486
  • [26] Unsupervised speaker adaptation for robust speech recognition in real environments
    Yamade, S
    Baba, A
    Yoshikawa, S
    Lee, A
    Saruwatari, H
    Shikano, K
    ELECTRONICS AND COMMUNICATIONS IN JAPAN PART II-ELECTRONICS, 2005, 88 (08): : 30 - 41
  • [27] Unsupervised feature selection using multi-objective genetic algorithms for handwritten word recognition
    Morita, M
    Sabourin, R
    Bortolozzi, F
    Suen, CY
    SEVENTH INTERNATIONAL CONFERENCE ON DOCUMENT ANALYSIS AND RECOGNITION, VOLS I AND II, PROCEEDINGS, 2003, : 666 - 670
  • [28] Synthetic-to-Real Composite Semantic Segmentation in Additive Manufacturing
    Petsiuk, Aliaksei
    Singh, Harnoor
    Dadhwal, Himanshu
    Pearce, Joshua M.
    JOURNAL OF MANUFACTURING AND MATERIALS PROCESSING, 2024, 8 (02):
  • [29] A lexicon driven approach to handwritten word recognition for real-time applications
    Kim, G
    Govindaraju, V
    IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 1997, 19 (04) : 366 - 379
  • [30] Handwritten Farsi/Arabic word recognition
    Broumandnia, A.
    Shanbehzadeh, J.
    Nourani, M.
    2007 IEEE/ACS INTERNATIONAL CONFERENCE ON COMPUTER SYSTEMS AND APPLICATIONS, VOLS 1 AND 2, 2007, : 767 - +