Study of augmentations on historical manuscripts using TrOCR

Author

Erez Meoded, Mississippi State UniversityFollow

Issuing Body

Mississippi State University

Advisor

Chen, Jingdao

Committee Member

Rahimi, Shahram

Committee Member

Mittal, Sudip

Date of Degree

12-8-2023

Document Type

Graduate Thesis - Open Access

Major

Computer Science

Degree Name

Master of Science (M.S.)

College

James Worth Bagley College of Engineering

Department

Department of Computer Science and Engineering

Abstract

Historical manuscripts are an essential source of original content. For many reasons, it is hard to recognize these manuscripts as text. This thesis used a state-of-the-art Handwritten Text Recognizer, TrOCR, to recognize a 16^th-century manuscript. TrOCR uses a vision transformer to encode the input images and a language transformer to decode them back to text. We showed that carefully preprocessed images and designed augmentations can improve the performance of TrOCR. We suggest an ensemble of augmented models to achieve an even better performance.

Recommended Citation

Meoded, Erez, "Study of augmentations on historical manuscripts using TrOCR" (2023). Theses and Dissertations. 6034.
https://scholarsjunction.msstate.edu/td/6034