CCTV-Smartphone-AI-Monitoring

Why we included this project

Old Android phones that would otherwise sit in a drawer can become network camera nodes, streaming to a PC dashboard over your LAN and giving you live preview, local recording, and screenshots from a browser with no cloud involved. The architecture splits mobile capture from PC-side processing, and the AI path runs in two stages: a cheap motion-detection pass decides when to wake a heavier vision model, which then writes structured event records with risk grading. That keeps it useful both as a small home or office monitoring setup and as a testbed for visual data collection and analysis. One thing to plan for: the Android client and the Flask server are separate components that have to be deployed together, and multimodal inference usually expects an online model service.