niessner / Matterport

Matterport3D is a pretty awesome dataset for RGB-D machine learning tasks :)
https://niessner.github.io/Matterport/
MIT License
929 stars 153 forks source link

The post processing method for the scene generation #35

Open bareblackfoot opened 4 years ago

bareblackfoot commented 4 years ago

Hi! I collected a 3D dataset with a Matterport pro2 camera and since the dataset is collected at a library, moving people are collected in several scans. I bought the MatterPak to get a .obj file and transformed it into a .glb file to use it at the habitat simulator. Since the matterport3d dataset is so clean, I think you used post-processing for making the scene clean. If you did use the post-processing algorithm, could you let me know how you removed cluttered point clouds? Thanks :)

czha5168 commented 3 years ago

Same question here. Thanks!