Detect Provenance
Drop an image to look it up in the Ma’atara Provenance index. The image is fingerprinted on your device, and only the fingerprints are sent — your image never leaves the device. Matches are ranked by how much of the image overlaps with registered work.
The hashes that cross the wire are one-way perceptual digests rather than the source image. They are not anonymous: a service can compare them with candidate material, which is the purpose of this search. Reconstructing the original from a digest alone is computationally impractical, but the digest can reveal a match to already-known content.
Rate Limits
Anonymous limits are per IP. The 60/min tier requires a server-verified, audience-scoped DID assertion and is counted per DID.
Query Image
Drop or pick an image
PNG · JPEG · WebP — processed on-device
How Detect works
Zero-image search. The query image is decoded and fingerprinted entirely in your browser. Only the resulting fingerprints — a whole-image measurement, measurements of individual regions, and an optional descriptor that recognises the subject — cross the wire. Your original image bytes never leave the device.
Matching. Your fingerprints are compared against those of registered work. Scoring combines how many regions overlap, how closely the subject matches, whether the overlap is spatially coherent, and how confident that makes the result. Results are tiered: Exact → Strong → Likely → Possible → Weak.
Privacy boundary. The fingerprints are one-way: reconstructing your image from them is computationally impractical. They are not anonymous, though — comparing them against material someone already holds can confirm a match, and that is precisely what the index is for.
Endpoint: POST /api/search/fragment on the Ma’atara Provenance worker.
Anonymous requests are limited to 10 per minute per IP. Requests with a server-verified,
audience-scoped MaataraAssertion are limited to 60 per minute
per DID.