Magical Image Duplicate Detector
This is a simple Python program to detect duplicate of images in a directory.
python main.py --directory <directory>Where directory is the path of the folder you want to run the script on. And certainty is the percentage of, well, certainty, that the script will consider two images identical.
When running on the sample_images directory the expected behavior is that the script will open your default web browser and show a temporary HTML file with the images in a table of image => [similar images]. With these three examples of Krystal arts, from Star Fox, the output should be two rows, of images 7ca89435aee9a06c3c4879347093ece6 and 9b54005da1051123d6b5d81e5d8f3e1b, which are identical (except for the resolution, that is different).
Through PyPI, @ https://pypi.org/project/midd/1.1.2/:
pip install middMidd works by using Pillow to get your image file, converting it to grayscale, and then shrinking it to 17x16 pixels. With this grayscale tiny little map it builds a true/false value, like a signature for said image.
This is done to every image inside the specified folder, and then, when it is done with this heavy work, the script starts comparing the images. If you have a folder with 4.000 images then this could take a few minutes, since it will do a little more than 17 million comparisons.
7ca89435aee9a06c3c4879347093ece6.jpg is like this:
Then it is turned to grayscale:
Which is resized to:
And it is actually 10101001010010100010100101001010011010010100101001001001010010100101100101001010010110010100101001011101010010100101111101001010010111110100101001011111010010100101111101001010010111110100101001011111010010100101111101001010010111110100101001011111010010000101111101001001 after passing through build_map in conversion.py. This hash is 100% equal to the hash produced by 9b54005da1051123d6b5d81e5d8f3e1b.jpg.