Duplicate Image Finder
Find duplicate and visually similar photos locally with perceptual hashing.
How it works
- 1Drop a folder or a selection of images — they are read directly in your browser, never uploaded.
- 2Each image is scaled to 64×64 grayscale and hashed with a perceptual hash (dHash).
- 3Hashes are compared pairwise with Hamming distance; images within the threshold are grouped.
- 4Review each duplicate group, remove the copies you do not need, and export the list.
FAQ
Are my images uploaded to a server?
No. Hashing and comparison run entirely inside your browser. Your images never leave your device.
Will it find resized or re-compressed copies?
Yes, usually. Perceptual hashing is robust to resizing, small crops and JPEG re-compression. Heavily edited versions (color shifts, big crops) may fall outside the threshold — raise it to catch them.
Is this really AI?
It uses a classical computer-vision technique (perceptual hashing) rather than a neural network — no model download is needed, so it works instantly, even offline.
How many images can it handle?
Hashing runs in a Web Worker at roughly 20–50 images per second. A few thousand images take under a minute; comparisons are done incrementally as each image is hashed.