ImageFolder Function
Zhenxing-Dong opened this issue · 5 comments
Hello!
When I tried to modify the ImageFoloder function, I found that none of my actions could change it. I also tried to print some variables in the ImageFolder function, but that was also impossible.
I would really appreciate your help!
Thank you very much!
Did you install compressai from your clone compressai code, or from pypi (pip install compressai)?
I followed this link "#102 (comment)" to install compressai.
Is there any difference between the two approaches?
By the way, since I wanted to compress the grayscale image, I tried to modify the network input channel from three to one , but also failed.
Would need more details to debug this. Please follow the issue report template + which exact commands you use: training / eval?
If you followed pip install -e . that should take your edits into account.
As for grayscale, you would need to define a model that accepts monochrome images, current architectures and pretrained models expect 3 component images (see conv(3, N))
I have changed the conv(3, N) to conv(1, N). But when I input a grayscale image, it still is conv(3, N).
It seems that my attempts to modify any function are futile.
It seems that my attempts to modify any function are futile.
Then, as i said, it must be your environment. Please follow the template that comes with the bug report. Other users may help you if you provide the following:
Please copy and paste the output from python3 -m torch.utils.collect_env
- PyTorch / CompressAI Version (e.g., 1.0 / 0.4.0):
- OS (e.g., Linux):
- How you installed PyTorch / CompressAI (`pip`, source):
- Build command you used (if compiling from source):
- Python version:
- CUDA/cuDNN version:
- GPU models and configuration:
- Any other relevant information: