IndexError: index_fill_(): Expected dtype int64 for index.

[mister-raupe]

Hello, thank you for providing the code and the installation instructions! Unfortunately I have run into this problem when I try to run the Demo. I am really unexperienced and can't resolve the issue myself. So any help is appreciated :)

Windows 11 Cuda 11.6 Anaconda

the full error message:

[vrai17]

same

[LSpych]

Adding 'dtype=torch.int64' worked for me, as in this example from line 95 of mesh_util.py:

self.smpl_mano_vertex_mask = torch.zeros(self.smpl_verts.shape[0],).indexfill(0,torch.tensor(self.smpl_mano_vid, dtype=torch.int64), 1.0

[YuliangXiu]

@LSpych Thanks, pull request for this fix is welcome, and you will be listed as contributor as well.

[2601705041]

same problem

[2601705041]

Adding 'dtype=torch.int64' worked for me, as in this example from line 95 of mesh_util.py:

self.smpl_mano_vertex_mask = torch.zeros(self.smpl_verts.shape[0],).indexfill(0,torch.tensor(self.smpl_mano_vid, dtype=torch.int64), 1.0 it works for me,but then i type 'python -m apps.infer -cfg ./configs/econ.yaml -in_dir ./examples -out_dir ./results' there is nothing happened

[alexwang2013]

@LSpych thanks. this works for me on windows.

my env is the following: windows 10 CUDA 12.1 MiniConda