A full description and discussion of Milovision is available here.
Collaborators are more than welcome: just contact the author.
Milovision is, first, a pose estimation pipeline. It turns images of a known circular marker into estimates of the camera's position and rotation relative to that marker.
This project includes an OpenGL based simulator that generates images of markers in the virtual camera's field of view. The simulator serves to validate the model, and to evaluate and improve the performance of the pose estimation pipeline.
While many visual marker based pose estimation systems do exist, the motivation for this project arose from the lack of freely available implementations based on circular markers. Circular marker based methods offer some distinct advantages for the purpose of pose estimation by comparison with those that rely on square markers, not least due to their ability to accurately detect partially occluded ellipses.
Refer to the Implementation chapter of the report for a short user manual.
For the impatient: edit, make executable, and run the convenience script,
run, or type python main.py -h.