Wie currently are working on some promising Computer vision Algorithms in the field of SLAM, Deep Sparse Odometry, Deep Parralel Tracking and Mapping (DTAM) and Eulerian Video Magnification...

Problem

Besides some exceptions most Computer Vision Algorithms like the Built In in Google's Project Tango, PTAM or in many MatchMoving Programs PfTrack or 3D Equalizer, they all still rely on feature based Tracking. A pretty old Algorithm that might be well researched but has enormous troubles with pattern-like structures. We would like to approach other attempts based on Motion detection within images and thereby rebuilding paralax shifts. Machine Learning such as Deep Sparse Coding might help us...

Follow the Development on GitHub:

https://github.com/freshNfunky

...more coming soon.