zhengzerong / multiviewsmplifyx Goto Github PK

I think not to add other params,when our camera params have known

Do you mean SMPL or SMPL-X model? Why SMPL_FEMALE.pkl appear in your example directory? @ZhengZerong

Hi,

Thanks for providing the code. May I know which version you are using. I am getting the below error.

Processing: dataset_example/image_data/rp_dennis_posed_004/color/0306.jpg
Processing: dataset_example/image_data/rp_dennis_posed_004/color/0324.jpg
Processing: dataset_example/image_data/rp_dennis_posed_004/color/0342.jpg
./vposer/models
Found Trained Model: ./vposer/models/snapshots/TR00_E096.pt
Stage: 0%| | 0/4 [00:00<?, ?it/s]
Orientation: 0%| | 0/1 [00:00<?, ?it/s]
Traceback (most recent call last):
File "/home/registration/MultiviewSMPLifyX/main.py", line 275, in
main(**args)
File "/home/registration/MultiviewSMPLifyX/main.py", line 250, in main
fit_single_frame(img_list, keypoints_list,
File "/home/registration/MultiviewSMPLifyX/fit_single_frame.py", line 395, in fit_single_frame
final_loss_val = monitor.run_fitting(
File "/home/registration/MultiviewSMPLifyX/fitting.py", line 175, in run_fitting
loss = optimizer.step(closure)
File "/home/registration/.conda/envs/pamir_registration/lib/python3.9/site-packages/torch/optim/optimizer.py", line 88, in wrapper
return func(*args, **kwargs)
File "/home/registration/MultiviewSMPLifyX/optimizers/lbfgs_ls.py", line 281, in step
orig_loss = closure()
File "/home/registration/MultiviewSMPLifyX/fitting.py", line 294, in fitting_func
body_pose = vposer.decode(
File "./vposer/models/vposer_smpl.py", line 114, in decode
if output_type == 'aa': return VPoser.matrot2aa(Xout)
File "./vposer/models/vposer_smpl.py", line 152, in matrot2aa
pose = tgm.rotation_matrix_to_angle_axis(homogen_matrot).view(batch_size, 1, -1, 3).contiguous()
File "/home/registration/.conda/envs/pamir_registration/lib/python3.9/site-packages/torchgeometry/core/conversions.py", line 233, in rotation_matrix_to_angle_axis
quaternion = rotation_matrix_to_quaternion(rotation_matrix)
File "/home/registration/.conda/envs/pamir_registration/lib/python3.9/site-packages/torchgeometry/core/conversions.py", line 302, in rotation_matrix_to_quaternion
mask_c1 = mask_d2 * (1 - mask_d0_d1)
File "/home/registration/.conda/envs/pamir_registration/lib/python3.9/site-packages/torch/_tensor.py", line 544, in rsub
return _C._VariableFunctions.rsub(self, other)
RuntimeError: Subtraction, the - operator, with a bool tensor is not supported. If you are trying to invert a mask, use the ~ or logical_not() operator instead.

Thanks for the help,
Sai

When I use handface paramter for smpl model fitting, there is an error tensor size 118 vs tensor size 25 for loss compute. how can I solve this problem ?