Göm meny

Learnable Camera Motion Models

This project runs 2015-2018.
The project runs for four years, and is funded by The Swedish Research Council (VR).

New: We are offering a masters thesis project on Machine Learning for Camera Based Trail Mapping. We have previously done video stabilization under difficult cases, see e.g. this video, and are now extending this to a full structure-from-motion system, that will handle video from e.g. a MTB mounted camera. The project involves 3D geometry, machine learning and opportunities for a scientific publication, and to attend an internatinal conference. Contact us for details!


The principal investigator for this project is Per-Erik Forssén.
The project also employs the PhD student Hannes Ovrén.

Research Topic

In this project, we develop continuous-time camera motion models that can be adapted to specific situations through learning. Such models are needed in video stabilisation and rectification on mobile platforms. They are also useful for control of motorised gimbals that mechanically stabilise the camera aim. They are also useful in 3D reconstruction from rolling shutter video.

The project is a continuation of the VGS project.


Most software developed in this project has been released as open source.
  • The Kontiki toolbox for continuous-time structure-from-motion.
  • The Crisp toolbox for camera-IMU self-calibration.



Senast uppdaterad: 2019-01-15