Thesis Details

Zvyšování konzistence v datových sadách pro rozpoznávání textu

Bachelor's Thesis Student: Tvarožný Matúš Academic Year: 2021/2022 Supervisor: Kišš Martin, Ing.
English title
Improving Consistency in Text Recognition Datasets
Language
Czech
Abstract

This work is concerned with increasing the consistency of datasets for text recognition. This paper describes the problems that cause the inconsistency and then presents solutions to eliminate it. The effect of the properties of the polygons defining the text line boundaries and hence how the modified version of the dataset, which is composed of ideal text line variants, affected the accuracy of the model is investigated. Further, the work focuses on detecting and then removing or modifying text lines whose ground truth transcription does not match the actual text they contain. Experimentation showed that removing the visual inconsistency on the training set did not have a significant effect on the trained model, but modifying the test set improved the OCR accuracy of the model by 1.1\% CER. By modifying the dataset so that it did not contain mutually inconsistent pairs of recognized text and the corresponding ground truth, the model improved by a maximum of only 0.2\% CER after re-training. The main finding of this work is, above all, the proven beneficial effect of removing inconsistencies on test suites, thanks to which it is possible to determine a more realistic error rate of the OCR model.

Keywords

text recognition, OCR, HTR, neural networks, NN, convolutional neural networks, CNN, recurrent neural networks, RNN, sequence to sequence, se2seq, CTC, consistency, datasets

Department
Degree Programme
Files
Status
defended, grade C
Date
14 June 2022
Reviewer
Committee
Zemčík Pavel, prof. Dr. Ing. (DCGM FIT BUT), předseda
Burget Lukáš, doc. Ing., Ph.D. (DCGM FIT BUT), člen
Holík Lukáš, doc. Mgr., Ph.D. (DITS FIT BUT), člen
Martínek Tomáš, Ing., Ph.D. (DCSY FIT BUT), člen
Matoušek Petr, doc. Ing., Ph.D., M.A. (DIFS FIT BUT), člen
Citation
TVAROŽNÝ, Matúš. Zvyšování konzistence v datových sadách pro rozpoznávání textu. Brno, 2022. Bachelor's Thesis. Brno University of Technology, Faculty of Information Technology. 2022-06-14. Supervised by Kišš Martin. Available from: https://www.fit.vut.cz/study/thesis/25170/
BibTeX
@bachelorsthesis{FITBT25170,
    author = "Mat\'{u}\v{s} Tvaro\v{z}n\'{y}",
    type = "Bachelor's thesis",
    title = "Zvy\v{s}ov\'{a}n\'{i} konzistence v datov\'{y}ch sad\'{a}ch pro rozpozn\'{a}v\'{a}n\'{i} textu",
    school = "Brno University of Technology, Faculty of Information Technology",
    year = 2022,
    location = "Brno, CZ",
    language = "czech",
    url = "https://www.fit.vut.cz/study/thesis/25170/"
}
Back to top