Francis Toupin

Since the deformation starts from an ellipsoid, it cannot generalize to meshs that aren't from the same genus, meaning that meshs with holes in it like the bike won't be...

Apart from the fact that the code is set to use images of dimension 224x224, there were no mentions of pre-processing in the paper or in the codebase.

Here is some information : https://github.com/walsvid/Pixel2MeshPlusPlus/issues/7#issuecomment-590088239 They are basically the pose parameters of the 3d model in front of the camera.

Hey, I'm not a maintainer of this project but I've looked at the source code quite a bit. It really depends if you want to always use the same number...

From where do you run the python script? Because tf_nndistance_so.so is loaded from a relative path. You should run the script directly from the project's root. If this isn't the...

If libcudart 9.0 can't be found, I believe it is because you are not using CUDA 9.0. Could you maybe give me the given path for CUDA (that you put...

Right now, the code is written so that it uses 3 images. You could "hack" it and pass 3 times the same image, it would work as all the statistics...

The Chamfer loss function doesn't require the two meshes to have the same number of vertices, so you could use that as a start. In general, Tensorflow doesn't care if...

@emres You can also simply add the GraphViz's bin folder to the Windows PATH variable without needing to do it in the Python script.

@klueska I'm starting from the tensorflow/tensorflow:2.7.0-gpu image, I tried following this blog and so I added the following lines at the start of my file (after FROM tensorflow...) RUN apt-get...