Closed royinblr closed 2 years ago
With model.compute_hash(method='phash', hash_size=16)
I was able to detect some duplicates but it seems that certain images require some pre-processing because a black or white border is present or the image is stretched. This makes a huge difference in the hash.
Example:
send me a pm if you want this functionality to be added, and we can discuss what's possible.
Not all the duplicates are detected.