Palette

A SvelteKit + FastAPI web app for plant and fruit classification with integrated color palette extraction.

Features

• Plant Identification: Upload one or more images to identify plant species using the PlantNet API.
• LLM Boost: Automatically boost low-confidence results (<80% confidence) using OpenAI GPT-4 Vision for image captioning.
• Color Palette Extraction: Extract dominant colors from uploaded images to create custom palettes.
• Theme Testing: Apply extracted colors to test areas (header, sidebar, content) in real-time.
• Palette Export: Download palettes in JSON or CSV formats.

Tech Stack

• Frontend: SvelteKit (TypeScript, Tailwind optional)
• Backend: FastAPI
• External APIs: PlantNet API, OpenAI GPT-4 Vision

Project Structure

Palette/
├─ palette/               # SvelteKit frontend
│  ├─ src/
│  │  ├─ routes/+page.svelte  # Main UI
│  │  └─ lib/color/extract.ts # Color extraction logic
│  └─ ...
└─ server/                # FastAPI backend
   ├─ main.py             # PlantNet + OpenAI Vision endpoints
   └─ ...

Setup

Prerequisites

• Node.js 18+

• Python 3.10+

• Git

• API keys for:

  • PlantNet (PLANTNET_KEY)
  • OpenAI GPT-4 Vision (OPENAI_API_KEY)

Installation

# Clone repo
git clone https://github.com/<your-username>/<your-repo>.git
cd Palette

# Frontend setup
cd palette
npm install

# Backend setup
cd ../server
python -m venv .venv
source .venv/bin/activate   # Windows: .venv\Scripts\activate
pip install -r requirements.txt

Environment Variables

Create a .env file in the server folder with:

PLANTNET_KEY=your_plantnet_api_key
OPENAI_API_KEY=your_openai_api_key

Running Locally

Start FastAPI backend:

cd server
uvicorn main:app --reload

Start SvelteKit frontend:

cd palette
npm run dev

The app will be available at http://localhost:5173 (frontend) and http://localhost:8000 (backend).

Deployment

• Frontend: Vercel, Netlify, or any SvelteKit-compatible host.
• Backend: Render, Railway, or any FastAPI-compatible platform.

License

MIT License. See LICENSE for details.