shurans / sscnet

Semantic Scene Completion from a Single Depth Image
http://sscnet.cs.princeton.edu/
339 stars 91 forks source link

Question regarding the benchmark #29

Open xcyan opened 6 years ago

xcyan commented 6 years ago

Dear SSCNET authors:

It looks like "benchmark/SUNCG_train_sceneId.mat" contains only ~6K scene ids.

Is there any specific reason not using the entire SUNCG dataset? We would appreciate if you can clarify or provide the filtering criterion (or code). Thanks!

Best, Xinchen