---- datatemplateentry publication ----
template            : publications:display_template
title               : Determining the Orientation of Low Resolution Images of a De-Bruijn Tracking Pattern with a CNN
date_date           : 2022-08-07
template            : publications:display_template
authors_            : [[people:andreas_schmid|Andreas Schmid]], [[people:raphael_wimmer|Raphael Wimmer]], Stefan Lippl
epub_url            : https://epub.uni-regensburg.de/cgi/users/home?screen=EPrint%3A%3AView&eprintid=53134
publisher_url       : https://dl.acm.org/doi/abs/10.1145/3532719.3543259
pdf_url             : https://epub.uni-regensburg.de/53134/1/dottrack_rotation_authorversion.pdf
bibtex_url          : https://epub.uni-regensburg.de/cgi/export/eprint/53134/BibTeX/epub-eprint-53134.bib
video_url           : 
doi                 : 10.1145/3532719.3543259
photo_img           : :projects:dottrack:dottrack_rotation.jpg
short-description   : We trained a deep neural network to determine the orientation of a tracking pattern in sensor images with extremely low resolution.
abstract            : Inside-out optical 2D tracking of tangible objects on a surface oftentimes uses a high-resolution pattern printed on the surface. While De-Bruijn-torus patterns offer maximum information density, their orientation must be known to decode them. Determining the orientation is challenging for patterns with very fine details; traditional algorithms, such as Hough Lines, do not work reliably. We show that a convolutional neural network can reliably determine the orientation of quasi-random bitmaps with 6 × 6 pixels per block within 36 × 36 pixel images taken by a mouse sensor. Mean error rate is below 2°. Furthermore, our model outperformed Hough Lines in a test with arbitrarily rotated low-resolution rectangles. This implies that CNN-based rotation-detection might also be applicable for more general use cases.
published-in        : ACM SIGGRAPH 2022 Posters
project             : dottrack
----