Automatic-Document-Scanner - not working

[orcohen9826]

Im runing the file "simple-scan.py" with python . I am using it to process some images that in the inputs folder, but the output images do not match the ones in the outputs folder. They look almost the same as the original images. I don't understand what you are doing differently to get the correct results.

[brmarkus]

Can you provide more information, @orcohen9826 please?

Are you talking about "https://github.com/spmallick/learnopencv/tree/master/Automatic-Document-Scanner" or "https://github.com/spmallick/learnopencv/tree/master/Document-Scanner-Custom-Semantic-Segmentation-using-PyTorch-DeepLabV3"?

Have you looked closer into the blog under "https://learnopencv.com/automatic-document-scanner-using-opencv/"?

In which step of the different phases do you see differences?

Have you done any changes to the code or the pictures?

Could it be due to the used OS, versions of Python, OpenCV, image-codecs? Can you comment on the used OS, Python, OpenCV, please?
How have you installed your environment, "natively", in a container, in a virt-env?

Which inputs - and especially which output do you get?

[brmarkus]

Now I just tried simple-scan.py under MS-Win10, using Python v3.8.8 - tested in a virt-env.
I needed to modify requirements.txt - to use opencv-contrib-python instead of opencv-contrib-python-headless.

Also needed to create a folder inputs under the folder outputs.

Using the sample input scanned-form.jpg - then I get the following (downscaled and cut screenshot):

What output do you get @orcohen9826 ?