Visual-Pattern Detection in Historical Documents

16
Not scheduled
20m
Von-Melle-Park 4

Von-Melle-Park 4

Poster

Description

Automatic pattern detection has become increasingly important for scholars in the humanities as the number of manuscripts that have been digitised has grown. Nevertheless, this task can be challenging for state-of-the-art computer vision approaches due to the lack of annotations and the small size of many patterns, which can be smaller than 0.1% of the image size. Therefore, we developed a training-free approach in order to overcome the first challenge, and a training-based approach in order to overcome the second challenge. Both approaches have been evaluated using research data from digitised manuscripts, and both yielded state-of-the-art results.

Keywords

Visual-Pattern Detection
Object Detection
Historical Documents Analysis

Author

Dr Hussein Mohammed (Universität Hamburg / The Cluster of Excellence: UWA)

Presentation materials

There are no materials yet.