Yang7879 / 3D-BoNet

🔥3D-BoNet in Tensorflow (NeurIPS 2019, Spotlight)

Home Page:https://arxiv.org/abs/1906.01140

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

The meaning of pc?

zhaoyu19920930 opened this issue · comments

Hi,Yang. I am reading the helper_data_s3dis.py, I have some confusion about the data contents.
Here are the code:
fin = h5py.File(file_path, 'r') coords = fin['coords'][block_id] points = fin['points'][block_id] semIns_labels = fin['labels'][block_id] pc = np.concatenate([coords, points[:, 3:9]], axis=-1)
I can not understand the ' pc = np.concatenate([coords, points[:, 3:9]], axis=-1)'
The coords mean the coordinates of the point clouds, and the semIns_labels are the semantices and instance labels of the point clouds, while the points means what? I know it's shape is (4096, 9).

could you please explain my confusion? Thanks

Hi @zhaoyu19920930 , the fin['coords'][block_id] represents the x-y-z of each point in the GLOBAL coordinate of a room (not a block), while the fin['points'][block_id][:, 0:3] represents the x-y-z of each point in the LOCAL coordinate of a block (not the room), the fin['points'][block_id][:, 3:6] represents the r-g-b of each point, the fin['points'][block_id][:, 6:9] represents the normalized x-y-z within the entire room.

The pc = np.concatenate([coords, points[:, 3:9]], axis=-1) takes out the global x-y-z and the remaining color / normalized x-y-z. Simply uncomment the lines below and visualize these data:

## if u need to visulize data, uncomment the following lines

why do u use global x-y-z and normalized x-y-z together(they are similar), and can i just use x-y-z without rgb(maybe not work?), wish for your answer