← Back to all updates
Tatuga School Update: Meet "FacePicker" — Transform Your Classroom with Live AI-Powered Webcam Student Selection! 🎬🤖
June 26, 2026

Tatuga School Update: Meet "FacePicker" — Transform Your Classroom with Live AI-Powered Webcam Student Selection! 🎬🤖

Move beyond traditional name wheels and bring a futuristic, gamified spark to your physical classroom! We are thrilled to launch FacePicker, a fully self-contained, fullscreen live camera picker powered by on-device AI. Point your webcam at the classroom, and let the AI track, spin, and spotlight your students in real time!

Authority Current Behavior & Refinements:

  • 👥 Classroom-Scale Person Detection: Reversing our early ideas for face-only models, FacePicker utilizes a powerful MediaPipe ObjectDetector customized for full-body tracking. It easily detects students sitting in the very back rows or caught at steep angles where standard facial landmarks vanish.
  • 👦🏻 Smart Head-Region Cropping: While the backend tracks full body profiles, our math engine runs a custom positioning filter that isolates the top-center slice (head region). The selection reticles and winning close-ups stay perfectly framed on the students' faces.
  • 🎚️ Adaptive Live Tracking: No static freezes here! The target boundaries and tracking vectors stay fully active through the entire 3-second deceleration phase. If students drift or jump in excitement during the raffle, the tracking grid moves fluidly alongside them.
  • 🎬 Cinematic Winner Reveal: At the winning microsecond, the active frame captures a high-resolution portrait of the chosen student. Driven by framer-motion, the cropped portrait animates smoothly to the dead center of the screen, complete with a vibrant canvas-confetti blast and crowd-cheering acoustics.
  • ⚙️ Google-Meet-Style Camera Settings: Built directly with external classroom environments in mind. If you deploy multi-camera setups, a dedicated gear overlay opens a quick settings popover listing all connected video inputs with live checkmark synchronization.
  • 🔐 Absolute Device-Level Isolation: Engineered with privacy as a foundational constraint. The entire architecture executes client-side via WebAssembly. No data is recorded, no names are registered, no facial recognition matching occurs, and zero video packets ever leave the host machine.