pqhieu / jsis3d

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

About preparing my own dataset #24

Closed lin1061991611 closed 2 years ago

lin1061991611 commented 3 years ago

hello,i am confued about the blocks in room_to_blocks in prepare_h5.py , what should i set up the threshold , and how dose cells work?

pqhieu commented 3 years ago

So usually a room will be split into different blocks, where each block is determined by size, with stride is the distance between one block to the next. The threshold depends on how dense your point cloud is and the maximum number of points in your training (in this case it is 4096).

Hope this helps.