423A35C7/pytorch3d
1
0
Fork 0
mirror of https://github.com/facebookresearch/pytorch3d.git synced 2025-12-21 14:50:36 +08:00
Code Issues Packages Projects Releases Wiki Activity

vert_align for Pointclouds object

Browse Source 
Reviewed By: gkioxari

Differential Revision: D21088730

fbshipit-source-id: f8c125ac8c8009d45712ae63237ca64acf1faf45
This commit is contained in:
Jeremy Reizenstein
2020-04-17 10:35:45 -07:00
committed by Facebook GitHub Bot
parent e19df58766
commit f25af96959
2 changed files with 61 additions and 20 deletions
Download Patch File Download Diff File Expand all files Collapse all files

27
pytorch3d/ops/vert_align.py
View File

@@ -25,10 +25,10 @@ def vert_align(
feats: FloatTensor of shape (N, C, H, W) representing image features
from which to sample or a list of features each with potentially
different C, H or W dimensions.
verts: FloatTensor of shape (N, V, 3) or an object (e.g. Meshes) with
'verts_padded' as an attribute giving the (x, y, z) vertex positions
for which to sample. (x, y) verts should be normalized such that
(-1, -1) corresponds to top-left and (+1, +1) to bottom-right
verts: FloatTensor of shape (N, V, 3) or an object (e.g. Meshes or Pointclouds)
with `verts_padded' or `points_padded' as an attribute giving the (x, y, z)
vertex positions for which to sample. (x, y) verts should be normalized such
that (-1, -1) corresponds to top-left and (+1, +1) to bottom-right
location in the input feature map.
return_packed: (bool) Indicates whether to return packed features
interp_mode: (str) Specifies how to interpolate features.
@@ -44,13 +44,11 @@ def vert_align(
resolution agnostic. Default: ``True``
Returns:
feats_sampled: FloatTensor of shape (N, V, C) giving sampled features for
each vertex. If feats is a list, we return concatentated
features in axis=2 of shape (N, V, sum(C_n)) where
C_n = feats[n].shape[1]. If return_packed = True, the
features are transformed to a packed representation
of shape (sum(V), C)
feats_sampled: FloatTensor of shape (N, V, C) giving sampled features for each
vertex. If feats is a list, we return concatentated features in axis=2 of
shape (N, V, sum(C_n)) where C_n = feats[n].shape[1].
If return_packed = True, the features are transformed to a packed
representation of shape (sum(V), C)
"""
if torch.is_tensor(verts):
if verts.dim() != 3:
@@ -58,8 +56,13 @@ def vert_align(
grid = verts
elif hasattr(verts, "verts_padded"):
grid = verts.verts_padded()
elif hasattr(verts, "points_padded"):
grid = verts.points_padded()
else:
raise ValueError("verts must be a tensor or have a `verts_padded` attribute")
raise ValueError(
"verts must be a tensor or have a "
+ "`points_padded' or`verts_padded` attribute."
)
grid = grid[:, None, :, :2] # (N, 1, V, 2)