search

pvnieo / GeomFmaps_pytorch

A minimalist pytorch implementation of: "Deep Geometric Functional Maps: Robust Feature Learning for Shape Correspondence"
MIT License
24 stars 2 forks source link

About Scape Dataset Ground-truth #2

Closed cainiaoshidai closed 3 years ago

cainiaoshidai commented 3 years ago

Hello, Thanks, I have got my matches on Scape dataset and now I want to calculate the geodetic error, but I don't know how to use the file of [.vts]. Just like you write in the Readme, the file of .vts is the ground truth maps, but I don't know how to get maps around any pairs shapes, just like the map mesh060.off to mesh061 et al. Looking forward to your reply, thanks.

Best.

pvnieo commented 3 years ago

Hi, The .vts files are ground truth map to a template mesh, so shape_1.vts is the p2p from the template to shape_1. Given a shape_1 and shape_2, and a predicted p2p map from shape_2 to shape_1 T21, the geodesic error is:

mean_geod_error = geodesic_matrix_shape1[T21[vts_shape2], vts_shape1].mean()
pvnieo commented 3 years ago

closing the issue, feel free to re-open it if you have further questions!