While taking EE364b: Convex Optimization II with Stephen Boyd, Ivan Papusha decided to do the following project:
... explores several large-scale methods for solving the Robust PCA problem via Principal Component Pursuit (PCP), a convex optimization problem, and demonstrates its use in background extraction tasks in video and point-cloud LIDAR data.
The idea is to perform background subtraction on a video stream. The PCP problem admits natural decomposition into Alternating Direction Method of Multipliers (ADMM, the course notes on ADMM are here) form. Here is the result: ‘‘Fast Automatic Background Extraction via Robust PCA’’. The poster is here. The matlab implementation is here.
Thanks Ivan for sharing.
Liked this entry ? subscribe to the Nuit Blanche feed, there's more where that came from
2 comments:
3 years later, the Matlab code is here (the link in the post is now broken) :
http://www.cds.caltech.edu/~ipapusha/code/pcp_admm.m
thanks Ronan !
Cheers,
Igoe.
Post a Comment