Awesome
About
demo - MATLAB code for our CVPR 2020 paper:
"Globally Optimal Contrast Maximisation for Event-based Motion Estimation", CVPR 2020, pdf
Daqi Liu, Álvaro Parra and Tat-jun Chin.
Description
demo.m runs CMGD [1] and our CMBnB and (contrast maximization with conjugate gradient and our impelmentation) on a subsequence from dynamic with about 19,000 events. To execute the demo just run demo.m in MATLAB. The script will:
- show the input stream and the event image without motion compensation,
- run CMBnB and CMGD, and
- plot motion compensated images for CMBnB and CMGD.
The expected runtime of the demo is less than 1 minute on a standard desktop PC.
Demo and code have been tested under
- Ubuntu 18.04
- MATLAB R2019a
- GCC v7
[1] Guillermo Gallego, Henri Rebecq, and Davide Scaramuzza. A unifying contrast maximization framework for event cameras, with applications to motion, depth, and optical flow estimation. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 3867–3876, 2018
Dependencies
eigen3 library,This library is included.
MATLAB global optimization toolbox
Compilation and Run
Binaries for Linux (Ubuntu) are supplied. To compile new binaries, please run compile.m in MATLAB. Code runs on full sequence of event is coming soon
Support
If you have any questions/bugs to report, please feel free to contact the author
License
For a closed-source version of CMBNB (e.g., for commercial purposes), please contact the author.
For an academic use of CMBNB, please cite Daqi Liu, Álvaro Parra and Tat-jun Chin, Globally Optimal Contrast Maximisation for Event-based Motion Estimation, CVPR 2020.