AI-RecogniX — Real-Time Face Recognition

Real-time face recognition system with multi-angle enrollment (Face ID–style), live recognition overlay, and voice alerts. Supports IP camera integration via RTSP tunnel.

PythonFastAPIASP.NET CoreReactdlibPostgreSQLDocker

Live demo Source

AI-RecogniX — Real-Time Face Recognition

What it does

Key features

Multi-angle face enrollment (front, left, right, up, down) — Face ID–style onboarding
Live webcam recognition with real-time identity badge overlay (green = authorized, red = unknown)
Voice alerts: "Authorized Access, [Name] is detected" or "Unauthorized Access!"
Subject management — add, view, and delete enrolled persons via dashboard
HOG-based dlib face detector with adaptive resize for close-up/high-res feeds
IP camera integration via RTSP tunnel (Hikvision NVR behind CGNAT using frp + go2rtc)
Fault-tolerant absent-frame detection — badge clears automatically when person leaves frame
Dockerized full-stack deployment (Python AI service + ASP.NET Core API + React frontend)

Under the hood

Architecture

AI ServicePython + dlib + face_recognition — HOG detection, 128-d embeddings, cosine matching

APIASP.NET Core Web API — enrollment, recognition, subject CRUD, EF Core + PostgreSQL

FrontendReact SPA — webcam capture, enrollment wizard, Live Recognize, Subjects management

DatabasePostgreSQL — subjects with serialized face encodings; EF Core migrations

IP Camerafrp reverse tunnel (CGNAT) → go2rtc on VPS → WebRTC stream to browser

DeploymentDocker Compose — ai, api, frontend, db services; Ubuntu VPS

Gallery

Screenshots

AI-RecogniX — Real-Time Face Recognition screenshot 1

AI-RecogniX — Real-Time Face Recognition screenshot 2

Interested in work like this?

Get in touch