AttributeError("'Tensor' object has no attribute 'pred_uvd_jts'")

[modaresimr]

Thank you for your great work. When i want to use demo_3d_inference.py it shows the following error: AttributeError("'Tensor' object has no attribute 'pred_uvd_jts'") at line 259

All the entries in this dictionary have this problem. Would you please help me?

smpl_output = {
    'pred_uvd_jts': pose_output.pred_uvd_jts.cpu()[new_ids],
    'maxvals': pose_output.maxvals.cpu()[new_ids],
    'transl': pose_output.transl.cpu()[new_ids],
    'pred_vertices': pose_output.pred_vertices.cpu()[new_ids],
}

[modaresimr]

More information: The complete command:

>python3 scripts/demo_3d_inference.py --cfg configs/coco/resnet/256x192_res50_lr1e-3_1x.yaml --checkpoint pretrained_models/fast_res50_256x192.pth --indir examples/demo/ --save_img
site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and will be removed in 0.15, please use 'weights' instead.
  warnings.warn(
site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and will be removed in 0.15. The current behavior is equivalent to passing `weights=ResNet50_Weights.IMAGENET1K_V1`. You can also use `weights=ResNet50_Weights.DEFAULT` to get the most up-to-date weights.
  warnings.warn(msg)
Loading pose model from pretrained_models/fast_res50_256x192.pth...
Loading YOLO model..
  0%|                                                                                                        | 0/3 [00:04<?, ?it/s]
AttributeError("'Tensor' object has no attribute 'pred_uvd_jts'")
An error as above occurs when processing the images, please check it

With the other configs:

>python3 scripts/demo_3d_inference.py --cfg configs/smpl/256x192_adam_lr1e-3-res34_smpl_24_3d_base_2x_mix.yaml --checkpoint pretrained_models/pretrained_w_cam.pth --indir examples/demo/ --save_img 
site-packages/torchvision/models/_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and will be removed in 0.15, please use 'weights' instead.
  warnings.warn(
site-packages/torchvision/models/_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and will be removed in 0.15. The current behavior is equivalent to passing `weights=ResNet34_Weights.IMAGENET1K_V1`. You can also use `weights=ResNet34_Weights.DEFAULT` to get the most up-to-date weights.
  warnings.warn(msg)
Loading pose model from pretrained_models/pretrained_w_cam.pth...
Loading YOLO model..
  0%|                                                                                                        | 0/3 [00:04<?, ?it/s]
_LinAlgError('cusolver error: CUSOLVER_STATUS_EXECUTION_FAILED, when calling `cusolverDnSgesvdjBatched( handle, jobz, m, n, A, lda, S, U, ldu, V, ldv, static_cast<float*>(dataPtr.get()), lwork, info, params, batchSize)`. This error may appear if the input matrix contains NaN.')
An error as above occurs when processing the images, please check it

[Fang-Haoshu]

Hi，the first command is not correct since you used the coco config. You should use the smpl config for 3d inference. In the second one, it seems the pytorch meet a cusolver error. What pytorch version are you using? Can you try the 2d inference to see if the environment works?