Errors in fitting.py

[Shruti Chakraborty]

[Shruti Chakraborty]

Hi! I am running the code as follows:

python smplifyx/main.py --config cfg_files/fit_smpl.yaml --data_folder Data --output_folder Output --visualize=True --model_folder Models --vposer_ckpt V02_05 --part_segm_fn smplx_parts_segm.pkl --use_cuda=False

I have changed interpretation = False in the .yaml file. I do not have cuda and want to see the results I can get from running on CPU.

Now I am seeing some errors in the fitting.py file in the lines 74 to 80:

body_pose = vposer.decode( pose_embedding, output_type='aa').view(1, -1) if use_vposer else None if use_vposer and model_type == 'smpl': wrist_pose = torch.zeros([body_pose.shape[0], 6], dtype=body_pose.dtype, device=body_pose.device) body_pose = torch.cat([body_pose, wrist_pose], dim=1)

I did some debugging and printed body_pose and it turns out body_pose is a dictionary which looks like this :

{'pose_body': tensor([[[-1.6974e-01, 3.9069e-02, 9.6282e-02], [-2.3981e-01, -5.0767e-02, -1.1809e-01], [ 2.4283e-01, 7.0910e-03, 9.4865e-03], [ 4.0049e-01, -2.3190e-02, -8.2887e-02], [ 4.7860e-01, -3.5693e-02, 4.6262e-02], [ 3.4491e-02, -4.0437e-02, 4.2832e-03], [-2.0321e-01, 2.0849e-01, 4.3582e-02], [-2.0967e-01, -1.5414e-01, 4.9889e-02], [ 3.9330e-02, 1.6497e-03, 5.0509e-03], [-1.6907e-04, 6.9093e-05, 9.9793e-06], [-4.4900e-05, 3.2197e-05, -2.7058e-05], [ 1.2475e-02, 2.0093e-02, -3.1115e-02], [-7.2320e-04, -5.3594e-02, -2.6114e-01], [ 1.2968e-02, 1.2447e-01, 2.2587e-01], [-7.8002e-02, -1.3200e-02, 4.9869e-02], [ 1.0644e-01, -2.0751e-01, -7.2850e-01], [ 1.4538e-01, 2.3604e-01, 7.4197e-01], [ 9.2566e-02, -9.1464e-01, 1.3255e-01], [ 9.8511e-02, 9.6942e-01, -2.3718e-01], [-3.0695e-01, -2.3050e-01, 8.9209e-02], [-2.2371e-01, 1.9913e-01, -4.0717e-02]]]), 'pose_body_matrot': tensor([[[ 9.9462e-01, -9.8952e-02, 3.0667e-02, 9.2343e-02, 9.8102e-01, 1.7049e-01, -4.6956e-02, -1.6674e-01, 9.8488e-01], [ 9.9179e-01, 1.2269e-01, -3.6070e-02, -1.1059e-01, 9.6449e-01, 2.3984e-01, 6.4216e-02, -2.3389e-01, 9.7014e-01], [ 9.9993e-01, -8.5366e-03, 8.1675e-03, 1.0250e-02, 9.7062e-01, -2.4041e-01, -5.8752e-03, 2.4048e-01, 9.7064e-01], [ 9.9635e-01, 7.6009e-02, -3.8914e-02, -8.5168e-02, 9.1753e-01, -3.8844e-01, 6.1802e-03, 3.9033e-01, 9.2065e-01], [ 9.9833e-01, -5.2867e-02, -2.3469e-02, 3.6113e-02, 8.8662e-01, -4.6108e-01, 4.5184e-02, 4.5946e-01, 8.8705e-01], [ 9.9917e-01, -4.9784e-03, -4.0344e-02, 3.5840e-03, 9.9940e-01, -3.4561e-02, 4.0492e-02, 3.4388e-02, 9.9859e-01], [ 9.7748e-01, -6.3987e-02, 2.0110e-01, 2.1925e-02, 9.7856e-01, 2.0479e-01, -2.0989e-01, -1.9577e-01, 9.5792e-01], [ 9.8695e-01, -3.3242e-02, -1.5755e-01, 6.5373e-02, 9.7691e-01, 2.0341e-01, 1.4715e-01, -2.1105e-01, 9.6634e-01], [ 9.9999e-01, -5.0172e-03, 1.7486e-03, 5.0820e-03, 9.9921e-01, -3.9316e-02, -1.5500e-03, 3.9324e-02, 9.9923e-01], [ 1.0000e+00, -9.9833e-06, 6.9092e-05, 9.9753e-06, 1.0000e+00, 1.6907e-04, -6.9093e-05, -1.6907e-04, 1.0000e+00], [ 1.0000e+00, 2.7058e-05, 3.2198e-05, -2.7057e-05, 1.0000e+00, 4.4900e-05, -3.2196e-05, -4.4901e-05, 1.0000e+00], [ 9.9931e-01, 3.1232e-02, 1.9894e-02, -3.0982e-02, 9.9944e-01, -1.2785e-02, -2.0282e-02, 1.2160e-02, 9.9972e-01], [ 9.6468e-01, 2.5808e-01, -5.2868e-02, -2.5804e-01, 9.6610e-01, 7.6711e-03, 5.3055e-02, 6.2418e-03, 9.9857e-01], [ 9.6693e-01, -2.2256e-01, 1.2455e-01, 2.2417e-01, 9.7455e-01, 1.1553e-03, -1.2164e-01, 2.6803e-02, 9.9221e-01], [ 9.9867e-01, -4.9282e-02, -1.5125e-02, 5.0311e-02, 9.9572e-01, 7.7560e-02, 1.1238e-02, -7.8218e-02, 9.9687e-01], [ 7.2683e-01, 6.4899e-01, -2.2478e-01, -6.7003e-01, 7.4194e-01, -2.4389e-02, 1.5094e-01, 1.6833e-01, 9.7410e-01], [ 7.1240e-01, -6.5051e-01, 2.6329e-01, 6.8306e-01, 7.2881e-01, -4.7561e-02, -1.6095e-01, 2.1373e-01, 9.6354e-01], [ 6.0277e-01, -1.5367e-01, -7.8298e-01, 7.4919e-02, 9.8784e-01, -1.3620e-01, 7.9439e-01, 2.3437e-02, 6.0695e-01], [ 5.4235e-01, 2.4326e-01, 8.0416e-01, -1.5550e-01, 9.6969e-01, -1.8846e-01, -8.2564e-01, -2.2838e-02, 5.6374e-01], [ 9.6985e-01, -5.1997e-02, -2.3810e-01, 1.2184e-01, 9.4957e-01, 2.8891e-01, 2.1107e-01, -3.0921e-01, 9.2728e-01], [ 9.7950e-01, 1.7995e-02, 2.0064e-01, -6.2205e-02, 9.7434e-01, 2.1629e-01, -1.9160e-01, -2.2434e-01, 9.5549e-01]]])}

And it is giving me TypeError in the vposer.decode function which says decode() got an unexpected keyword argument 'output_type':

I removed the output_type and then also got an error for .view() part of the argument which shows an AttributeError:

So I also removed that and now the changed code looked like the following:

body_pose = vposer.decode( pose_embedding) if use_vposer else None if use_vposer and model_type == 'smpl': wrist_pose = torch.zeros([body_pose.shape[0], 6], dtype=body_pose.dtype, device=body_pose.device) body_pose = torch.cat([body_pose, wrist_pose], dim=1)

BUT this still gives errors as the wrist_pose = torch.zeros([body_pose.shape[0], 6],
dtype=body_pose.dtype,
device=body_pose.device)
seems to want bodu_pose to be a numpy array but is a dictionary as I showed above. The dictionary body_pose also does not have any attribute called "device". So I changed the code to :

body_pose = vposer.decode( pose_embedding) if use_vposer else None if use_vposer and model_type == 'smpl': print(body_pose) wrist_pose = torch.zeros([body_pose['pose_body'].shape[0], 6], dtype=body_pose['pose_body'].dtype) print(wrist_pose) body_pose = torch.cat([body_pose['pose_body'], wrist_pose], dim=1)

However, at this point I've made multiple changes to the code and in the last line the tensor dimensions do not agree.

Can someone tell me what is going on and where I am going wrong?
Am I using an old version of the code?

Thanks a lot @vchoutas @geopavlakos @dimtzionas @MichaelJBlack

[YixuannnL]

I have the same problem, have you figured it out?

@vchoutas @geopavlakos @dimtzionas @MichaelJBlack @shrutic12