🥗 Food Calories & Proteins Analyzer

An AI-powered nutrition assistant that analyzes food images to detect food items, calculate calories & protein content, and answer nutrition-related questions. Built with LangGraph, Google Gemini, Nutritionix API, Wikipedia, and Streamlit.

✨ Features

Food Recognition – Upload an image, and Gemini identifies visible food items.
Calorie & Protein Calculation – Nutritionix API provides accurate nutrition facts.
Interactive Q&A – Ask follow-up questions about the analysis (e.g., “How much protein is in the rice?”).
Wikipedia Integration – Learn nutrition benefits of foods via Wikipedia.
Conversation Memory – Keeps track of analysis and follow-up questions using MemorySaver.
Streamlit UI – Simple, user-friendly interface for interaction.
MCP Server (Render) – Nutrition tools are served via a FastAPI MCP server hosted on Render.

🛠 Tech Stack

LangGraph – Workflow orchestration & memory
Google Gemini – Vision + LLM for food recognition & reasoning
Nutritionix API – Nutrition facts (calories, protein, etc.)
Wikipedia API – Extra nutrition knowledge
Streamlit – Frontend for interaction
FastAPI + MCP – Backend tool serving (deployed on Render)

📂 Project Structure

.devcontainer/
├── devcontainer.json       # Main config file
├── backend/                # Backend API & services
│   ├── requirements.txt    # Backend dependencies
│   ├── server.py           # API server
│
├── frontend/               # Streamlit-based frontend
│   ├── app.py              # Main app
│   ├── client.py           # Connects to backend
│   ├── requirements.txt    # Frontend dependencies
│
├── .env.example            # Example env file
├── .gitignore              # Git ignore rules
├── .python-version         # Python version
├── pyproject.toml          # Project metadata
├── README.md               # Project documentation
├── requirements.txt        # Root dependencies
└── uv.lock                 # Lockfile (reproducible builds with uv)

⚡ Setup Instructions

1️⃣ Clone the repository

git clone https://github.com/sunilpowar5/LangGraph-MCP-Nutrition-Analyzer.git
cd LangGraph-MCP-Nutrition-Analyzer

2️⃣ Create virtual environment & install dependencies

python -m venv .venv
# Activate venv
.venv\Scripts\activate      # Windows
source .venv/bin/activate   # Mac/Linux

pip install -r requirements.txt

3️⃣ Add environment variables

Create a .env file in the root directory:

GEMINI_API_KEY=your_google_gemini_api_key
NUTRITIONIX_APP_ID=your_nutritionix_app_id
NUTRITIONIX_API_KEY=your_nutritionix_api_key
LANGCHAIN_API_KEY=your_langchain_api_key

(You can use .env.example as a template)

4️⃣ Run the app (Streamlit frontend)

streamlit run frontend/app.py

5️⃣ (Optional) Run backend MCP server

uvicorn backend.server:app --host 0.0.0.0 --port 8000

Usage

Upload a food image (jpeg/jpg/png). The system detects food items → Nutritionix fetches calories & protein info.

Ask follow-up questions like:

“Is this meal healthy?”
“What are the benefits of the meal?”

Start a new session anytime.

📌 Example Workflow

Upload image → 🍌 + 🥛 detected

Nutritionix fetch →

Banana: 105 calories, 1.3g protein
Milk: 103 calories, 8g protein
Total: 208 calories, 9.3g protein

Ask: “What are the benefits of milk?” → Answer fetched from Wikipedia.

🚀 Roadmap

Support multi-food image segmentation (bounding boxes).
Add more nutrition APIs for wider coverage.
Deploy on Streamlit Cloud / AWS.
Multi-user history & persistent storage.

🤝 Contributing

Contributions, issues, and feature requests are welcome! Feel free to open a PR or report a bug.

LangGraph Nutrition Analyzer