Related: 231220-1232 GBIF iNaturalist plantNet duplicates

KilianB/JImageHash: Perceptual image hashing library used to match similar images does hashes based on image content, not bytes (a la SHA1 and friends)

Hashing Algorithms · KilianB/JImageHash Wiki is a cool visual explanation of the algos involved.

Kind of Like That - The Hacker Factor Blog is a benchmark thing, TL;DR

• aHash is very quick but many FP
• dHash just as quick but better

One of the comments suggest running a quick one with many FPs and then a slower one on the problematic detected images.