how to run demo on CPU

[sychou11]

I wanted to run demo on CPU so I tried
python demo.py --vid_file demo.mp4 --gpu -1
However, it turned out:
RuntimeError: Attempting to deserialize object on a CUDA device but torch.cuda.is_available() is False. If you are running on a CPU-only machine, please use torch.load with map_location=torch.device('cpu') to map your storages to the CPU.

Then I tried to fix code in demo.py in line 108 and line 127.
But it still couldn't work.

[hongsukchoi]

Did you also change the code here?

[sychou11]

Did you also change the code here?
It works, thanks!

[AhmedBourouis]

i did the same thing and i changed the code as follow:
demo.py {
line 108: ckpt = torch.load(pretrained_file,map_location=torch.device('cpu'))
line 127: checkpoint = torch.load(osp.join(BASE_DATA_DIR, 'spin_model_checkpoint.pth.tar'),map_location=torch.device('cpu'))
}
tcmr.py {
line 134: pretrained_dict = torch.load(pretrained,map_location=torch.device('cpu'))['model']
}
but I'm still getting this error:

AssertionError:
Found no NVIDIA driver on your system. Please check that you
have an NVIDIA GPU and installed a driver from
http://www.nvidia.com/Download/index.aspx

[hongsukchoi]

@AhmedBourouis

Hmm, it seems there is no problem. Would you trying eliminating this line?

[AhmedBourouis]

@hongsukchoi
thank you for your answer but i still have the same error even after eliminating that line (line 374 in demo.py)
can i run it in google colab ?

[hongsukchoi]

No, I haven't prepared google colab codes.

I recommend you follow this instructions to run on CPU. I did the same thing and was able to run TCMR on CPU.
In demo.py

L55 : device = torch.device('cpu')
L108: ckpt = torch.load(pretrained_file,map_location=torch.device('cpu'))
L115: model.regressor.smpl = SMPL(
        SMPL_MODEL_DIR,
        batch_size=64,
        create_transl=False,
        gender=gender
    ) #.cuda()
L127: checkpoint = torch.load(osp.join(BASE_DATA_DIR, 'spin_model_checkpoint.pth.tar'),map_location=torch.device('cpu'))

In tcmr.py

L134: pretrained_dict = torch.load(pretrained, map_location=torch.device('cpu'))['model']