some problem about the figure about votenet HOT 7 CLOSED

Hi @WangZhouTao . Happy to learn you liked the paper. Indeed we used meshlab. Fun fact: you can make the boxes look nicer by turning them to quadmesh (under FIlter -> remeshing)

from votenet.

and the point color is coded according to the depth of the points.

as Or said, for remeshing in MeshLab you can select Filters->Polygon and Quad Mesh->Turn into Quad-dominant mesh.

from votenet.

@charlesq34 Thanks very much for sharing. So the color is computed in advance before imported into MeshLab, or colorized by some function in MeshLab?

from votenet.

Hi @NUAAXQ No problem.

The color is written in advance into the PLY file. You can refer to the write_ply_rgb here

We used matplotlib's colormap.

from votenet.

Thank you for your reply,

from votenet.

@WangZhouTao Were you able to get the figure ?

from votenet.

Hi @NUAAXQ No problem.

The color is written in advance into the PLY file. You can refer to the write_ply_rgb here

votenet/utils/pc_util.py

Line 229 in 257b8d8

def write_ply_rgb(points, colors, out_filename, num_classes=None):

We used matplotlib's colormap.

Hello, can I add color information during training and see the color of point cloud during visualization:
CUDA_VISIBLE_DEVICES=0 python train.py --dataset sunrgbd --log_dir log_sunrgbd --use_color

from votenet.