Deploy and Setup
- download the dataset, which is a single .mat file.
- modify the “Consts.m” file. In this file, we need to setup the correct path. This includes:
- download and install SPAMS framework. Download and build the package, paste the absolute path to “build” folder to the Const.m
- download and install gurobi optimization package.
Start to Run
There are three entry points to the code package. The first is the run_pipeline_segmentation.m. This only generate the segmentation results. The other two are the run_pipeline_support_inference_gt.m and run_pipeline_support_inference_seg.m. These two do support inference based on the manually labeled segments and the segments generated by the segmentation procedure respectively.
Issue in running in 2013a
The original package is written and compiled in 2011a, there are some changes between the versions of Matlab. The code will fail in an error “not positive semi-definite”. We need to create a new function and copy the content in “princomp.m” into it, then change this line
varargout{1} = pca(varargin{1},’Algorithm’,’eig’,’Economy’,fEconomy);
to use the ‘svd’ algorithm. Then change all the places that call princomp to call this new funciton.