get_xyz_vertex_embedding does not always use the device

[matejsuchanek]

The code that creates the embedding map for "smpl_27554" mesh does not put tensors to the provided device, unlike all other possible meshes (the else branch):

detectron2/projects/DensePose/densepose/vis/densepose_outputs_vertex.py

Lines 22 to 37 in 94113be

	def get_xyz_vertex_embedding(mesh_name: str, device: torch.device):
	if mesh_name == "smpl_27554":
	embed_path = PathManager.get_local_path(
	"https://dl.fbaipublicfiles.com/densepose/data/cse/mds_d=256.npy"
	)
	embed_map, _ = np.load(embed_path, allow_pickle=True)
	embed_map = torch.tensor(embed_map).float()[:, 0]
	embed_map -= embed_map.min()
	embed_map /= embed_map.max()
	else:
	mesh = create_mesh(mesh_name, device)
	embed_map = mesh.vertices.sum(dim=1)
	embed_map -= embed_map.min()
	embed_map /= embed_map.max()
	embed_map = embed_map**2
	return embed_map

[github-actions]

You've chosen to report an unexpected problem or bug. Unless you already know the root cause of it, please include details about it by filling the issue template.
The following information is missing: "Instructions To Reproduce the Issue and Full Logs"; "Your Environment";

[github-actions]

Requested information was not provided in 7 days, so we're closing this issue.

Please open new issue if information becomes available. Otherwise, use github discussions for free-form discussions.