GhostFaceMPC runs face recognition under multi-party computation
Original title: GhostFaceMPC: Privacy-Preserving Face Recognition with Multi-Party Computation
Authors quantize GhostFaceNet so extraction, storage, and matching stay encrypted and call the design the first end-to-end MPC system of its kind.
A cs.CR paper presents GhostFaceMPC, a privacy-preserving face recognition system that evaluates a deep convolutional extractor and matching under secure multi-party computation. The authors say raw images and biometric embeddings are not exposed during feature extraction, storage, or matching. They introduce an MPC-friendly quantization of GhostFaceNet and circuit optimizations intended to make recognition practical. Code is published at github.com/crypto-unict/GhostFaceMPC.
Unconfirmed: The claim of being the first end-to-end MPC face-recognition system is the authors' assertion.
- The paper targets unauthorized access to biometric face data.
- GhostFaceNet is quantized to fit multi-party computation circuits.
- Matching as well as feature extraction runs under MPC.
- Source code is at github.com/crypto-unict/GhostFaceMPC.
- Products
- GhostFaceMPC
- AI models
- GhostFaceNet
From the original article
1 min read at arxiv.orgFacial recognition has become central to modern authentication systems, security frameworks, and personalized services. However, its widespread adoption is hindered by severe privacy risks, particularly surrounding unauthorized access to sensitive biometric data.
By Davide Carnemolla · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.