Launched this week
FaceUnlock
Hands-free Mac unlock using your face - private, local, free
4 followers
Hands-free Mac unlock using your face - private, local, free
4 followers
FaceUnlock brings Face ID-style unlocking to macOS. Enroll your face in seconds, then just glance at your camera to unlock - no extra hardware needed, just your existing FaceTime camera. Built on ArcFace, running fully on the Apple Neural Engine, with zero network calls and zero telemetry. Your password and face data stay encrypted on-device, gated by Touch ID. Free and source-available. Ideal for anyone tired of Touch ID failing when your hands are full or you're leaning back at your desk.
Hey Product Hunt!
I'm the developer behind FaceUnlock. I built this because Touch ID kept failing me in the exact moments I needed it most - hands full carrying stuff, or when I've got my Mac connected to an external monitor and it's turned away from me, with Touch ID just a bit too far to reach comfortably. I wanted the hands-free convenience of Windows Hello, but nothing comparable existed for the Mac.
How it works: FaceUnlock enrolls your face (7 quick poses) using ArcFace, converted to Core ML and running entirely on the Apple Neural Engine. When your Mac wakes from sleep, FaceUnlock automatically triggers a scan on the lock screen - no need to do anything else. You can also manually trigger it anytime by pressing Space or Return.
A few things I cared about while building this:
Zero network calls, zero telemetry - everything happens on-device
Your password and face embeddings are AES-GCM encrypted, gated behind Touch ID
Passive liveness detection so a static photo can't spoof it
It's a convenience layer, not a security replacement - your Mac password is still the real lock underneath
It's free, and the app code is open source (the face model itself is used under InsightFace's non-commercial license, which I've documented clearly in the repo). I'm a student building this in my spare time, so it's not notarized yet - that requires a $99/year Apple Developer membership I haven't been able to swing, hence the one-line "xattr" workaround on first install. There's a Ko-fi link if anyone wants to help me get there faster.
Would genuinely love feedback - especially on the enrollment flow, the security model, or edge cases I might not have thought of (lighting conditions, look-alikes, etc.). Happy to answer anything!