Skip to content
Zester
All products

Applied Vision

FoodLens Agent

A restaurant photo enhancement agent: a dish photo goes in, marketing-ready imagery comes out.

Domain
Applied Interfaces
Status
Prototype
Stack
Google GeminiFastAPIReactVite

6

Style presets

21

Pytest tests, TDD throughout

The problem

Small restaurants photograph their own food and it shows.

Marketing imagery is a professional service priced for chains. An independent restaurant has a phone, a dish, and no realistic path from one to the other.

How it works

  1. 01

    Take the photo that exists

    A phone photo of the dish is the input. No lighting rig, no reshoot.

  2. 02

    Enhance against cuisine

    Google Gemini applies cuisine-specific prompting across six style presets rather than one generic filter.

  3. 03

    Export where it will be posted

    Output is Instagram-ready, which is the actual destination for this image.

Architecture

  1. 01

    Input

    Dish photograph uploaded from the restaurant

  2. 02

    Generation

    Google Gemini with cuisine-specific prompts across 6 style presets

  3. 03

    Service

    FastAPI backend, 21 pytest tests written test-first

  4. 04

    Surface

    React and Vite interface with Instagram-ready export

Capabilities

  • Dish photo enhancement
  • Six cuisine-aware style presets
  • Instagram-ready export
  • Test-driven FastAPI service

Have an operation this should be pointed at?

Solve a Real Problem