Similar or Duplicate images aren't recognized at all

elisemercury / Duplicate-Image-Finder

difPy - Python package for finding duplicate or similar images within folders

https://difpy.readthedocs.io

MIT License

448 stars 67 forks source link

Similar or Duplicate images aren't recognized at all #62

Closed bayoleems closed 1 year ago

elisemercury commented 1 year ago

Hi @bayoleems

Can you please provide a bit more detail to your issue? With no information, we cannot relate to the issue you are having.

Thank you and best, Elise

bayoleems commented 1 year ago

Hi @elisemercury I've got similar and duplicate images in the directory but the package keeps returning 0 values for duplicate/similar Images

elisemercury commented 1 year ago

Hi @bayoleems,

Thanks for clarifying! I'll be happy to try and find the issue you are running into - for that, could you please answer the following questions for me?

Which version of difPy are you running?
Which file types are the images you are comparing?
Could you please provide me with the output of search.stats?
If you would like difPy algorithm to find similar images, I recommend to set similarity to 'low'. Could you please try this and check the output? When similarity is set to 'normal' it will search for duplicate images.
Any specific reason you are using px_size=10 instead of the default value 50?

Thank you and all the best, Elise

bayoleems commented 1 year ago

Hi @elisemercury

I'm using the recent version from the repo and comparing .jpg image files.

from the search.stats json there's a number of invalid files which I think might be a reason for the issue. What does it mean by invalid file?

Lastly I used px value as 10 cause I thought it would be more precise in similarity matching

I still get the same thing even after changing the similarity to low and px to 50.

Thank you.

elisemercury commented 1 year ago

Hi @bayoleems,

It seems that multiple people have the same issue as you with the new version of difPy. If possible, could you please share one (or more) of the images with me, so that I can test what the issue is related to? That would be of great help. You can share the image with me via elisejlandman@hotmail.com. I will work on fixing the issue ASAP!

The issue you experienced is likely related to the Pillow package. Pillow made some adjustments with versions >9.1.0 that can affect difPy in the background, depending on which one is installed on your PC. The issue has now been fixed with difPy v3.0.4. Please update it and let me know if the issue is fixed from your end as well.

Thank you and best regards, Elise