search

pqhieu / jsis3d

[CVPR'19] JSIS3D: Joint Semantic-Instance Segmentation of 3D Point Clouds
https://pqhieu.com/research/cvpr19/
MIT License
176 stars 36 forks source link

sceneNN #30

Closed lllmmmwww closed 2 years ago

lllmmmwww commented 3 years ago

Thank you for your contribution!I have a question. If I use a dataset like sence NN, Do I need to turn room into block in prepareh5. py？

QingWindIsStillTheWind commented 2 years ago

Have you solved this problem now

pqhieu commented 2 years ago

Thank you for your contribution!I have a question. If I use a dataset like sence NN, Do I need to turn room into block in prepareh5. py？

Yes. You can download the ply files from SceneNN, extract the point clouds and run through preprocessing.