Tanuj Jain
Tanuj Jain
Could you also update the relevant test cases please?
Thanks for the report, will make a note in the documentation.
Tested it on M1 max and it works. Don't have quick access to M1 unfortunately. Meanwhile, just using the space to collect resources for the fix: 1. https://stackoverflow.com/questions/65966969/why-does-march-native-not-work-on-apple-m1 2. https://github.com/grpc/grpc/issues/29262...
Could you please share more info? - imagededup version - OS - conda/pip version Also please try to stick to import formats as defined in the [documentation](https://idealo.github.io/imagededup/methods/hashing/).
Haven't explicitly tested it, but compression shouldn't be too much of a trouble, especially for CNN method. Maybe you could give it a go on a small hand-picked collection of...
Would be great if you could share some of your results.
@PetrochukM Did you manage to give it a go?
@PetrochukM So, this package didn't work for you? Also, could you share which software packages you ended up using?
I can reproduce it too. Will have a look in a bit. Meanwhile, you could make progress by using 'bktree' as the search method like below: `hasher.find_duplicates('/path/to/image/directory', search_method='bktree')` The search...
The culprit here is the `convert` function from `PIL` which is being used to convert your `RGBA` image to `RGB`. Both `cnn` as well as hashing methods rely on this...