This program searches a directory for potentially duplicate images using perceptual hashing algorithms.
In experiments, the program is able to find all 1,002 duplicate images (out of a dataset of 11,002 total images) in about 40 seconds. This efficiency is achieved through the use of BK-Trees, which allows fast hash comparison. For more information, see here.
Place the script in the folder containing the images.