search

mks0601 / I2L-MeshNet_RELEASE

Official PyTorch implementation of "I2L-MeshNet: Image-to-Lixel Prediction Network for Accurate 3D Human Pose and Mesh Estimation from a Single RGB Image", ECCV 2020
MIT License
710 stars 130 forks source link

Can't download pretrained models! #51

Closed Yarp96 closed 3 years ago

Yarp96 commented 3 years ago

Hi, Im trying to download the pretrained models from https://drive.google.com/drive/folders/14SJKZuCudli2xNN6pSh1bJepIZYlBfSE however im getting the error: The number of downloads was exceeded.

Is there any other way to download pretrained models?

Best regards,

mks0601 commented 3 years ago

Please read README more carefully. That issue is already addressed in here.

Yarp96 commented 3 years ago

Thanks for the fast answer! Unfortunatly even after copying the model to my own drive it is still saying that the number of downloads was exceeded.

mks0601 commented 3 years ago

Did you try download from your google drive file, not from the shared one?

Yarp96 commented 3 years ago

Hi again,

Copying to my drive did not work however when I download the whole folder from my shared drive it zipped it and I was able to download it!

Thank you for your time!

Yarp96 commented 3 years ago

Sorry to bother again, I have been able to run the demo by following your steps however it seems like the output isn't what it should be and im not sure why.

Here is a picture of the mesh on cloud compare: image output_mesh_lixel output_mesh_param rendered_mesh_param

I am using snapshot_8.pth, checked the bbox and it matches the person we want to predict on.

There is a warning when loading the model: UserWarning: The given NumPy array is not writeable, and PyTorch does not support non-writeable tensors. This means you can write to the underlying (supposedly non-writeable) NumPy array using the tensor. You may want to copy the array to protect its data or make it writeable before converting it to a tensor. This type of warning will be suppressed for the rest of this program. (Triggered internally at ..\torch\csrc\utils\tensor_numpy.cpp:141.) torch.Tensor(smpl_data['betas'].r).unsqueeze(0))

If you have any idea what could cause this let me know!

mks0601 commented 3 years ago

I've just downloaded the repo and ran demo. The results are below. Please double check you followed the instruction carefully.

rendered_mesh_param rendered_mesh_lixel

Yarp96 commented 3 years ago

Hi again,

Thanks for the fast answer but I found my mistake, I removed dataparallel thinking that It wouldnt affect the code but it actually messed up the predictions. Sorry for bothering!