~/work / face-recognition
futurescape/oneapp-facesproduction
Visitor Face Recognition
Recognizes returning visitors at society gates in ONEAPP, a society-management platform serving 50,000+ users. Built on face detection, embeddings and nearest-neighbour matching.
- face-detection
- face-embeddings
- similarity-search
01 / problem
Why it needed to exist
Security guards re-register the same visitors again and again. Gate entry should know who has been there before.
02 / what i built
The approach
01
Detect → embed → match
Faces are detected in the camera frame and converted to embeddings, then matched against known visitors with KNN / distance thresholds.
02
Production integration
Integrated into the live Flutter app alongside REST APIs, Firebase and native Android / iOS components.
03 / architecture
How data flows
- 00Cameragate device frame
- 01Detectlocate & crop face
- 02Embedface → vector
- 03KNN matchdistance to known visitors
- 04Decisionreturning visitor or new
04 / results
What it did in the real world
- Shipped in ONEAPP, a platform serving 50,000+ users