Giters

mrharicot / monodepth

Unsupervised single image depth prediction with CNNs

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

world coordinates

KunikaValecha opened this issue · comments

commented

I suppose, the depth produced from disparity map is in world coordinates. Then given calibration matrix, how can I generate coordinates in world frame?
My approach is to make homogeneous vector from (xpixel, ypixel) and then multiply it with inverse of calibration matrix. That will That will give us ratio of x/z, y/z. Further ratio * z(from disparity) will give xworld, yworld. Is this correct?