search

hkust-vgd / pointwise

Code for Pointwise Convolutional Neural Networks, CVPR 2018
http://pointwise.scenenn.net
MIT License
130 stars 26 forks source link

What are the attributes stored in each point? #17

Closed tranvnhan closed 5 years ago

tranvnhan commented 5 years ago

I could interpret the data structure of each .hdf5 file as follows:

where N denotes the number of blocks each contains 4096 points sampled from that particular block; 12 is the number of attributes stored in each point.

I know that the first three attributes are X,Y,Z coordinates. What about the rest of it? Do they imply anything?

And one more question. What tool or package did you use to visualize the point clouds? image

songuke commented 5 years ago

Regarding the format, according to this evaluation script, it should be xyz normal color global_xyz. https://github.com/scenenn/pointwise/blob/master/scene_seg/eval_scene_seg_scenenn.py The global_xyz is used for merging all the blocks into a global scene.

Thanks.

2019年8月18日(日) 11:03 Nhan notifications@github.com:

I could interpret the data structure of each .hdf5 file as follows:

-

data field: shape(N, 4096, 12)

label field: shape(N, 4096)

where N denotes the number of blocks each contains 4096 points sampled from that particular block; 12 is the number of attributes stored in each point.

I know that the first three attributes are X,Y,Z coordinates. What about the rest of it? Do they imply anything?

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/hkust-vgd/pointwise/issues/17?email_source=notifications&email_token=AABBPMR5JQTUE3N3WVYBR3LQFDC23A5CNFSM4IMRI7IKYY3PNVWWK3TUL52HS4DFUVEXG43VMWVGG33NNVSW45C7NFSM4HFZ73QQ, or mute the thread https://github.com/notifications/unsubscribe-auth/AABBPMVEC6J6FWTFDFNBER3QFDC23ANCNFSM4IMRI7IA .

tranvnhan commented 5 years ago

That should be clear enough.

Regarding the visualization, what package did you use to view the point clouds?

pqhieu commented 5 years ago

We used our in-house visualisation tool to view the point clouds. You can achieve the same results using Phong shading and screen space ambient occlusion.

Take a look at Open3D too.

Thanks.

tranvnhan commented 5 years ago

We used our in-house visualisation tool to view the point clouds. You can achieve the same results using Phong shading and screen space ambient occlusion.

Take a look at Open3D too.

Thanks.

Thank you very much.