SimpleStereo There are four main routines that can run individually. All of them take input dataset from ./imgs. For example, motorcycle dataset will be in ./imgs/motorcycle. Make sure you have im0.png im1.png im1E.png im1L.png calib.txt disp0.pfm for each dataset at its subfolder. And save the result at ./result. NCC_main.m - compute disparity using ncc patch matching SSD_main.m - compute disparity using ssd patch matching corr_main.m - compute disparity using raw cross corrletaion patch matching DP_main.m - compute disparity using dynamic programing To do quantitative evaluation, run quantitative_evaluation.m It will take the result in the result folder and compute the root mean square to the ground truth.