Progress

24 Jan 2014 - Finished convolutions scripts for segmentation, need to down
sample the image: 77 filters by H*w*3 is computational tasking.  To do: finish
downsampling: will try uniform sampling of points to convolve. 1000 points per
image should be more than enough I think. 

25 Jan 2014 - Downsampled by just randomly picking points out of the convolutions,
not super ideal since it would be better to avoid calculations on points that
are not used. Something for later potentially/going to try to get it up tonight
so that some training for the CRF can happen tomorrow. In between everything I plan
to write my interim report.