Scribe Vision
AI-powered whiteboard camera
Recognition·1st Place — Senior Design Pitch Day
This project was awarded first place among all senior design teams at the NC State College of Engineering startup pitch competition.
Problem we're solving
Many students have experienced frantically copying the board instead of actually listening and learning. Missing a lecture. Sitting too far back to read the content. These aren't minor inconveniences; they are barriers to education.
Solution: What is Scribe Vision?
A portable AI device that instantly converts all physical whiteboard content into structured, actionable digital notes.
Overview
Scribe Vision is a headless hardware device built on a Raspberry Pi 5. It automatically captures images of whiteboard lectures, uses an custom fine-tuned LLM to understand and transcribe the notes, and compiles them into cleanly formatted, downloadable LaTeX PDFs, no keyboard or monitor required.
The device broadcasts its own web application, controllable entirely from a smartphone or laptop.
How it works
Raspberry Pi 5
Records and captures whiteboard images autonomously
Flask App
Mobile-responsive control panel served from the Pi
LLM Transcribe
Images to subject-aware LaTeX via prompt-guided AI
Cloud Compile
Raw .tex compiled to PDF on a DigitalOcean droplet
Download
Formatted PDF available instantly in the web library
Key Capabilities
- Fully headless — operates without a keyboard, mouse, or monitor
- Auto-hotspot fallback for first-time network configuration
- TFT display shows a QR code linking directly to the web app
- Subject-aware transcription for math, chemistry, and physics notation
- Cloud-compiled LaTeX for fast, high-quality PDF output