search

alexsax / 2D-3D-Semantics

The data skeleton from Joint 2D-3D-Semantic Data for Indoor Scene Understanding
http://3dsemantics.stanford.edu
Apache License 2.0
464 stars 67 forks source link

About semantic labels #28

Closed ghost closed 5 years ago

ghost commented 5 years ago

Hi,

Thank you very much for opening a great dataset.

Can I ask some questions?

I engage in a research about semantic segmentation of 3D point clouds by using 2D semantic labels based on camera pose.

So, what I want to obtain ( and I don't still obtain) are:

  1. Ground truth of 3D semantic labels(may be semantic.mtl and semantic.obj? If so, Please let me know how to extract.)
  2. Do /data/semantic or semantic_pretty/camera*.png data mean 2D semantic labels? I want to know which object each label belongs to. Therefore, what I want is followed as: [120, 120, 0] -> wall [20, 40, 200] -> ceiling . . .

I'm very happy that you respond.

ustundag commented 5 years ago

Hi,

Regarding your second question, maybe you can look at this issue; #6

ir0 commented 5 years ago

Hi,

  1. In each area_#/3d folder, you can find the corresponding semantic.obj and semantic.mtl files. The encoding is the same as the one for the semantic images, and you can follow a similar process to the next point.
  2. As the previous comment pointed out, you can take a look at issue #6 for more information.
ghost commented 5 years ago

Dear researchers,

Thank you very much for your kind replies.

I succeeded to extract each pixel's class label.

Thank you very much !

alexsax commented 5 years ago

@AkiyoshiKurobe great! I'm closing this for now.

ghost commented 5 years ago

Thank you! But, I have one question about 3D semantic labels. Do you have any solutions on obtain 3D semantic labels?