Presentation
49. Determining the Orientation of Low Resolution Images of a De-Bruijn Tracking Pattern With a CNN
Contributors
Event Type
Poster
In Person
Full Conference Supporter
Full Conference
Exhibitor Experience
Exhibitor Additional Experience
Experience
Experience Plus
Exhibitor Additional Full Conference
Exhibitor Full Conference
Time
Location
DescriptionConventional computer vision algorithms for determining an image's rotation cease to work at low resolutions. We present a convolutional neural network that can predict the orientation of 36x36 pixel images of a tracking pattern based on a De-Bruijn torus in real time with a mean error rate of below 2°.
