Back to Projects
AI / Mobile Security

MTK AlertPro

An AI-powered mobile security app that transforms any IP camera into an intelligent surveillance system with on-device person/vehicle detection, instant push notifications, and alarm integration.

🚀

Project Overview

A privacy-first mobile security application built with React Native + Expo that turns any IP camera into a smart surveillance system. Runs TensorFlow.js with SSD MobileNet V2 directly on-device for real-time person and vehicle detection — zero cloud processing means zero latency on inference and complete data privacy. Features instant push notifications with configurable cooldowns, alarm sound playback with red alert mode, vibration patterns, and per-camera detection settings (person/vehicle toggles, sensitivity thresholds). Achieves under 2-second end-to-end alert latency. Connects to 80%+ of IP cameras via RTSP/HTTP streams with automatic reconnection and offline camera caching. Backend powered by Supabase with Row Level Security for all user data. Monetized via RevenueCat subscriptions.

Tech Stack

React NativeExpo SDK 52TypeScriptNativeWind 4Expo Router 4Zustand 5TanStack Query 5SupabaseTensorFlow.jsSSD MobileNet V2Firebase Cloud MessagingRevenueCatExpo Notifications

Key Features & Highlights

On-device AI detection — TensorFlow.js + SSD MobileNet V2 runs directly on phone for real-time person/vehicle detection with zero cloud dependency
Smart alert system with alarm integration — push notifications, configurable cooldowns, alarm sounds, red alert mode, and per-camera detection settings
Universal IP camera compatibility — 80%+ camera support via RTSP/HTTP streams with auto-reconnection and offline caching
Privacy-first architecture — all AI processing on-device, camera credentials encrypted in memory, Supabase RLS protecting all user data