I want to turn an ordinary wall mirror into a smart, non-touch display that recognises who is standing in front of it and responds automatically. The key challenge—and your main brief—is to design and implement accurate facial recognition so the mirror instantly identifies each registered user, then triggers their preferred lighting scene through a compatible system such as Philips Hue, Zigbee, or HomeKit. Here is what I need the finished solution to do: • Detect a face, match it to the stored profile, and greet the user within two seconds. • On successful recognition, call the lighting API with that person’s preset brightness, colour temperature, or scene. • Run everything locally (Raspberry Pi, Jetson Nano, or similar) so no cloud connection is required once the model is trained. • Display a clean, fullscreen interface on a non-touch HDMI display hidden behind two-way glass; the interface can be minimal—just the name and a subtle status icon—because interaction happens automatically. • Provide a simple admin script or GUI to enrol new faces, edit lighting preferences, and remove users. Deliverables 1. Annotated source code for the vision model, enrolment tool, and lighting integration. 2. Wiring diagram and parts list covering camera, display controller, and smart-bulb bridge. 3. Step-by-step setup guide so I can reproduce the build from a fresh SD card image. 4. A short test video or live demo confirming recognition accuracy and lighting control. This is a focused proof-of-concept; once it works reliably we can expand later into weather, voice, or health data, but for now I only care about seamless personalisation and automated lighting. If you have experience with OpenCV, TensorFlow-Lite or Face Recognition libraries and you enjoy bringing IoT ideas to life, I’d love to see your approach and timeline.