Question on generating the ground truth `w0`

zhou13 / nerd

NeRD: Neural 3D Reflection Symmetry Detector

MIT License

98 stars 20 forks source link

Question on generating the ground truth `w0` #3

Closed yanconglin closed 3 years ago

yanconglin commented 3 years ago

HI, Yichao,

I am having some trouble understading this line: https://github.com/zhou13/symmetrynet/blob/830bd9430c1750d31faffba60e18b075374c1b42/sym/datasets.py#L95

If I understand correctly, w0 is the the ground truth w. As you mentioned the shapenet objects are symmetrically aligned w.r.t Y-Z (x=0) plane, I suppose this line should be w0_ = LA.inv(RT).T @ np.array([1, 0, 0, 0]). Or am I completely wrong?

BTW, many congrats on your promotion! Impressive work! Looking forward to your reply.

Yancong

zhou13 commented 3 years ago

Hi Yancong,

Thanks for asking. The problem here is that ShapeNet releases in the obj format and it seems that obj wavefront specification only says things about "forward direction" and "upward direction" rather than xyz axes. Therefore, different software has different mapping for xyz for obj files: See the difference between blender and meshlab:

The implementation here uses the blender convention. We will update the paper accordingly to avoid confusion. Nice catch!

Best, Yichao Zhou

yanconglin commented 3 years ago

Thanks very much for your answer! I will look into the details. Issue closed for now.

njnj0314 commented 2 years ago

Hello, I recently read this paper, download the dataset encountered some problems, please ask you have this paper dataset?