Publikationen und Abschlussarbeiten am Lehrstuhl für Rechnernetze
Automation of Tedious Typing: Recognition of Handwritten Points on Exam Cover Sheets
Authors
Markus Brenneis
Fabian Mersch
Title
Automation of Tedious Typing: Recognition of Handwritten Points on Exam Cover Sheets
Published
In Proceedings,
Published in
Proceedings of International Conference on Information Technology and Applications
Abstract
Grading handwritten exams is a re-occurring task in education. Oftentimes, this means having grading tables with handwritten numbers on a cover sheet. Digitizing these tables is necessary for several purposes, including final grading. But copying all the information by hand to the right student record is very tedious. We developed a software solution for digitizing grading tables and recognizing the handwritten numbers. It uses 2d markers for identifying the relevant image section and a convolutional neural network for digit recognition. Tests involving two teachers have shown that this method is three times faster than the old manual method, although marker and digit recognition accuracy can be improved. We provide a new dataset containing over 11,000 handwritten digits and 500 grading tables to support the development of future algorithms.
