423A35C7/pytorch3d
1
0
Fork 0
mirror of https://github.com/facebookresearch/pytorch3d.git synced 2025-08-03 12:22:49 +08:00
Code Issues Packages Projects Releases Wiki Activity

Plotly subsampling fix

Browse Source 
Summary: When viewing two or more pointclouds in a single plot, we should be subsampling each one separately rather than subsampling their union.

Reviewed By: nikhilaravi

Differential Revision: D27010770

fbshipit-source-id: 3c7e04a6049edd39756047f985d5a82c2601b3a2
This commit is contained in:
Jeremy Reizenstein 2021-03-16 12:34:43 -07:00 committed by Facebook GitHub Bot
parent ff9c6612b4
commit 1b6182bac2
1 changed files with 15 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

20
pytorch3d/vis/plotly_vis.py
View File

@ -61,14 +61,14 @@ def plot_scene(
**kwargs, **kwargs,
): ):
""" """
Main function to visualize Meshes and Pointclouds. Main function to visualize Meshes, Cameras and Pointclouds.
Plots input Pointclouds, Meshes, and Cameras data into named subplots, Plots input Pointclouds, Meshes, and Cameras data into named subplots,
with named traces based on the dictionary keys. Cameras are with named traces based on the dictionary keys. Cameras are
rendered at the camera center location using a wireframe. rendered at the camera center location using a wireframe.
Args: Args:
plots: A dict containing subplot and trace names, plots: A dict containing subplot and trace names,
as well as the Meshes and Pointclouds objects to be rendered. as well as the Meshes, Cameras and Pointclouds objects to be rendered.
See below for examples of the format. See below for examples of the format.
viewpoint_cameras: an instance of a Cameras object providing a location viewpoint_cameras: an instance of a Cameras object providing a location
to view the plotly plot from. If the batch size is equal to view the plotly plot from. If the batch size is equal
@ -574,11 +574,21 @@ def _add_pointcloud_trace(
pointclouds = pointclouds.detach().cpu() pointclouds = pointclouds.detach().cpu()
verts = pointclouds.points_packed() verts = pointclouds.points_packed()
features = pointclouds.features_packed() features = pointclouds.features_packed()
total_points_count = max_points_per_pointcloud * len(pointclouds)
indices = None indices = None
if verts.shape[0] > total_points_count: if pointclouds.num_points_per_cloud().max() > max_points_per_pointcloud:
indices = np.random.choice(verts.shape[0], total_points_count, replace=False) start_index = 0
index_list = []
for num_points in pointclouds.num_points_per_cloud():
if num_points > max_points_per_pointcloud:
indices_cloud = np.random.choice(
num_points, max_points_per_pointcloud, replace=False
)
index_list.append(start_index + indices_cloud)
else:
index_list.append(start_index + np.arange(num_points))
start_index += num_points
indices = np.concatenate(index_list)
verts = verts[indices] verts = verts[indices]
color = None color = None