About installation
thanhdat77 opened this issue · 6 comments
I looked inside your 'setup.py' and your 'environment.yml' noticed that the opencv-python library is missing.
In your 'depthNet.py', I found that you are using absolute paths. I’ve changed all of them, but I still can't run the code.
Can you point out where i can fix it
Hi,
The path is just a placeholder. You should change them based on your needs. Try this checkpoints/DFF-DFV.tar. Also, if you are using Google Colab, it is possible that it is causing problems. The provided code and files has been tested, so the problem might be from your path, or google colab.
Have you thought of making a huggingface space? The people need a taste
Have you thought of making a huggingface space? The people need a taste
@kagevazquez Yes, we're planning to publish a Hugging Face Space soon! In the meantime, we've shared all the necessary codes and files to make everything easily accessible.
Hi @kagevazquez,
I tried running it locally and adjusted the path, but I’m still encountering the same issue. I also noticed that when I use os.environ["CUDA_VISIBLE_DEVICES"] = "2", it causes CUDA to become unavailable.
For more context, I'm using WSL with PyTorch 2.1.0 and CUDA 11.8.
Thank you for your contribution to this project. I'm excited to see how it evolves and to test your code on Hugging Face
@thanhdat77
About the `os.environ["CUDA_VISIBLE_DEVICES"] = "2". Thanks for catching it. This one is for multi-GPU training when you want to select a specific GPU to train or evaluate. I pushed a bug fix for removing this; it should fix your problem right now.
Regrading your error, how about you use absoulute paths?
I am closing this issue. Since I assume the problem has been solved.