Neuralangelo——（chamferdistance）

[xsl316]

Hello, mate. I came here after seeing your message on the Neuralangelo project and chose to leave a message on your github page because I didn't have your contact details. I saw that you inquired about Neuralangelo author accuracy evaluation, I would like to ask, did you succeed in evaluating the chamfer distance, and if so can you provide me with assistance? I am sorry if I am disturbing you.

[kf-zhang]

I have given up on reproducing the results of the paper. There are point cloud data in the DTU dataset (https://roboimagedata.compute.dtu.dk/?page_id=36), which should be usable for calculating the Chamfer distance.

[xsl316]

Thank you very much for your reply. For the calculation of chamfer distance, I can never reproduce the value in the paper, I know that you sent the GT point cloud of the connected DTU for the calculation of chamfer distance, but I would like to know the code that can accurately calculate the chamfer distance, do you have any research in this area?

[kf-zhang]

I remember that the coordinate systems of the point clouds in GT and the predicted point clouds are not aligned (you can confirm this by opening both point cloud files in Blender). It might require multiplying by a matrix to align the coordinates, but I wasn't able to find this matrix, so I didn't pursue it further. I hope this helps you.

[xsl316]

Thank you very much, my friend, and I wish you all the best!