The goal of the lab is to detect a binary map of an occlusion.
Download image sequence. Estimate 2D homography between the original photo and
the images (similarly as in the previous labs).
For each image (first for one, then for all of them) estimate the occlusion mask using graphcut.
Use the normalized cross-correlation to determine the unary terms and the Potts model for the binary term.
Use the following matlab function vgg_graph_maxflow.m
and its windows compiled mex.