torch_radon_cuda.forward() method returns tensors with all zero

[braynisboring]

Thank you for your excellent work on this project, but I ran into some problems and wanted to confirm with you.
I have followed your documentation as:

• git clone https://github.com/matteo-ronchetti/torch-radon.git
• cd torch-radon
• python setup.py install

After I run these commands，the terminal displays these commands

and I use python and import torch and import torch_radon_cuda,The program reports no errors.

But,when I use :

  x = torch.randn(2, 1, 512, 512).to("cuda")
  from torch_radon import Radon
  angeles = np.linspace(0, np.pi, 512, endpoint=False)
  radon = Radon(512,angeles)
  pro = radon.forward(x)

I found that the pro here is a tensor with all zero, I also converted this tensor into a PIL image to save, the picture is all black.
I want to know where is the problem.
Hope for your reply

[braynisboring]

At the same time, I noticed that the radon.forward() method is implemented through the torch_radon_cuda.forward() method of RadonForward(), but I found that there is noforward()method in torch_radon_cuda, is this normal, or am I doing something wrong with the installation.

The image is the entire content of torch_radon_cuda.py

[linfengyu77]

I have the same problem. Did you solve this problem ?

[braynisboring]

I have the same problem. Did you solve this problem ?

not yet..

[Love-Sunshine]

I have the same problem. Did you solve this problem ?

[amberhappy]

Same problem!

[Mucixia]

解决了吗

[Mucixia]

同样的问题！

请问你解决了吗