FIG. 006 / FACEID ONCHAINAn experiment in verifiable records. Overview.
FaceID Onchain explores how an image-processing pipeline can leave a verifiable record on Solana. The prototype combines local face detection and descriptor extraction with a cascading visual-search layer and a small Anchor attestation program.
A fingerprint of the input.
The Bun client uses face-api models to detect a face, align landmarks, and extract a 128-dimensional descriptor. A SHA-256 hash of that descriptor identifies the input record and derives its on-chain account.
A record that can be inspected.
The Anchor program stores the authority, descriptor hash, source URL, match flag, and timestamp. A verification command recomputes the input fingerprint and compares it with the stored account. This records the pipeline's result; it does not independently establish a person's identity.
The system at a glance.
- 01Extract a descriptor→
- 02Run visual search→
- 03Anchor the result→
- 04Inspect the record
Keep exploringHigh-Throughput GitHub Analytics & Viral Data Engine