FaceHunter is a face recognition module I built from scratch. Upload a photo — it finds a matching face in a database of 1.3 million in under a second. We sold it as a self-hosted license or a cloud API.
Three capabilities
The module does three things:
- Detect faces in any photo or video frame, returning pixel coordinates of face rectangles
- Compare two faces and get a similarity percentage
- Search a pre-built database of faces — 1.3 million records, 1 second response time
How we sold it
Two distribution models: install the module on your own server with a license key, or use our cloud API via HTTP. Either way, clients could test with a demo first to check speed and resource consumption.
Demo projects
We built a few fun demos to showcase what FaceHunter can do:
- FindCheck — upload a photo and search against a database of 1.3M faces from escort sites. Popular with dating platforms and marriage agencies
- Children.DNA — upload photos of mom, dad, and child to see who the kid looks more like. A viral demo that drove a lot of traffic
