FridgeChef

Take a photo of your fridge. Get three recipes back.

FridgeChef sends the photo to Google Cloud Vision for ingredient detection, lets you edit the result as chips, then asks Gemini 2.5 Flash for three complete recipes with ingredients, timings, and steps.

Requirements

• Python 3.11+
• A Google Cloud project with the Vision API enabled, plus a service-account JSON key
• A Gemini API key from Google AI Studio

Setup

1. Clone and enter the repo.
2. Create a virtualenv and install deps:

   python -m venv .venv
   # macOS/Linux: source .venv/bin/activate
   # Windows PowerShell: .venv\Scripts\Activate.ps1
   pip install -r backend/requirements.txt
   

3. Copy backend/.env.example to backend/.env and fill in both credentials.
4. Run the server:

   uvicorn backend.main:app --reload
   

5. Open http://localhost:8000.

Tests

pytest

17 tests covering the food-filter logic and both API routes (Vision and Gemini clients mocked).

Frontends

Two are available; pick whichever fits.

• Web (frontend/) — vanilla HTML/CSS/JS. Served automatically at http://localhost:8000 by the backend. No build step.
• Android (flutter_app/) — native Flutter app with camera + gallery picker. See flutter_app/README.md for Flutter SDK setup, emulator config, and how to run on a physical device.

Both talk to the same /api/detect and /api/recipes endpoints.

Architecture

See docs/superpowers/specs/2026-04-23-fridgechef-design.md for the full design.