search

huaaaliu / RGBX_Semantic_Segmentation

MIT License
316 stars 30 forks source link

S3DIS dataset #13

Closed OPradelle closed 2 years ago

OPradelle commented 2 years ago

Hi thanks for sharing your work.

I wanted to know how did you generate the 2D ground truth for the S3DIS dataset. I followed the original procedure from the dataset (https://github.com/alexsax/2D-3D-Semantics/issues/6) but i did not manage to reproduce the score presented in your paper.

Thanks for your answer

huaaaliu commented 2 years ago

Hi, thanks for your interest to our work. We does not use some particular methods to generate the 2D ground truth. We use the method mentioned in https://github.com/charlesCXK/Depth2HHA-python to generate the HHA image, please note the original depth format of the S3DIS and the camera parameters. Thanks