2026 · Web app
RootFacts
A browser app that recognizes vegetables through the camera and writes a fun fact about each one, with all its machine learning running on the device.
- Period
- –
- Client
- Personal project
- Role
- Sole developer, as a Dicoding course submission: implemented the camera, detection and fact-writing services and the app's flow in the course's React and Vite starter.
Problem
A Dicoding course asked for a web app that recognizes images and generates text entirely in the browser, with no backend and no API keys.
Outcome
The live app loads a Teachable Machine model that tells 18 vegetables apart in TensorFlow.js, counts a detection only at 70% confidence or more, and writes a fun fact about it with Transformers.js, all on the visitor's device.
Stack
- React
- Vite
- JavaScript
- TensorFlow.js
- Transformers.js
