This is a fork of chrdiller's implementation of the Chamfer Distance as a module for Pytorch. It is written as a custom C++/CUDA extension.

I modified it to also return the indices of the matching points, since it is common to need them as well.

As it is using pyTorch's JIT compilation, there are no additional prerequisite steps that have to be taken. Simply import the module as shown below; CUDA and C++ code will be compiled on the first run.

from chamfer_distance import ChamferDistance
chamfer_dist = ChamferDistance()

#...
# points and points_reconstructed are n_points x 3 matrices

dist1, dist2, idx1, idx2 = chamfer_dist(points, points_reconstructed)
loss = (torch.mean(dist1)) + (torch.mean(dist2))


#...