search

EdwardSmith1884 / 3D-IWGAN

A repository for the paper "Improved Adversarial Systems for 3D Object Generation and Reconstruction".
MIT License
195 stars 56 forks source link

The problem of Reconstruction of 3D Objects from Single Perspective Depth Scans #26

Closed caixialiu-bjut closed 3 years ago

caixialiu-bjut commented 4 years ago

Hi, thank you for your code. I have the problem about "Reconstruction of 3D Objects from Single Perspective Depth Scans". 1.In the 3D_generation.sh

convert to our format

python convert_shapenet10.py 3DShapeNets
python Kinect_setup.py chair

The data produced by above codes are from ShapeNet . However, the data in the paper show ModelNet, such as ' Example synthetic single perspective kinetic scans created by the authors, produced from the chair class of the ModelNet10 dataset.' in fig8. It's a contradiction.

2.I guess that Kinect-VAE-3D-IWGAN.py aims at fig8, and is it right? Could you share the code that converts Kinect RGB-D images (A Large Dataset of Object Scans dataset [33] ) to voxel representations(fig.10) ?

Thanks very much! Look forward to your reply.

EdwardSmith1884 commented 4 years ago

Hi,

  1. Yes the names are confusing here, if you download the data you will see that it is in fact ModelNet. The citation for ModelNet is weirdly enough : "3D ShapeNets: A Deep Representation for Volumetric Shapes", the download is correct however.

  2. Unfortunately I lost track of this code. It was never in any way automated to the degree that it could be shared or ready for public release.

Hope this helps,

Ed