TY - CHAP
T1 - One-shot Compositional Data Generation for Low Resource Handwritten Text Recognition
AU - Souibgui, Mohamed Ali
AU - Biten, Ali Furkan
AU - Dey, Sounak
AU - Fornes, Alicia
AU - Kessentini, Yousri
AU - Gomez, Lluis
AU - Karatzas, Dimosthenis
AU - Llados, Josep
N1 - Funding Information:
This work has been partially supported by the Swedish Research Council (grant 2018-06074, DECRYPT), the Spanish project RTI2018-095645-B-C21, the CERCA Program / Generalitat de Catalunya, the project PID2020-116298GB-I00, AGAUR project 2019PROD00090 (BeARS) and PhD scholarships from UAB (B18P0073).
Publisher Copyright:
© 2022 IEEE.
PY - 2022
Y1 - 2022
N2 - Low resource Handwritten Text Recognition (HTR) is a hard problem due to the scarce annotated data and the very limited linguistic information (dictionaries and language models). For example, in the case of historical ciphered manuscripts, which are usually written with invented alphabets to hide the message contents. Thus, in this paper we address this problem through a data generation technique based on Bayesian Program Learning (BPL). Contrary to traditional generation approaches, which require a huge amount of annotated images, our method is able to generate human-like handwriting using only one sample of each symbol in the alphabet. After generating symbols, we create synthetic lines to train state-of-the-art HTR architectures in a segmentation free fashion. Quantitative and qualitative analyses were carried out and confirm the effectiveness of the proposed method.
AB - Low resource Handwritten Text Recognition (HTR) is a hard problem due to the scarce annotated data and the very limited linguistic information (dictionaries and language models). For example, in the case of historical ciphered manuscripts, which are usually written with invented alphabets to hide the message contents. Thus, in this paper we address this problem through a data generation technique based on Bayesian Program Learning (BPL). Contrary to traditional generation approaches, which require a huge amount of annotated images, our method is able to generate human-like handwriting using only one sample of each symbol in the alphabet. After generating symbols, we create synthetic lines to train state-of-the-art HTR architectures in a segmentation free fashion. Quantitative and qualitative analyses were carried out and confirm the effectiveness of the proposed method.
KW - Document Analysis
UR - http://www.scopus.com/inward/record.url?scp=85126146250&partnerID=8YFLogxK
U2 - 10.1109/WACV51458.2022.00262
DO - 10.1109/WACV51458.2022.00262
M3 - Chapter
AN - SCOPUS:85126146250
T3 - Proceedings - 2022 IEEE/CVF Winter Conference on Applications of Computer Vision, WACV 2022
SP - 2563
EP - 2571
BT - Proceedings - 2022 IEEE/CVF Winter Conference on Applications of Computer Vision, WACV 2022
PB - Institute of Electrical and Electronics Engineers Inc.
ER -