EcoScan

EcoScan is a mobile app that enables to you track and analyze the ecological footprint of food products. By scanning barcodes on the go, EcoScan connects with a local Open Food Facts database and leverages intelligent AI agents to deliver detailed insights into the environmental impact of your daily choices.

For setting up your local environment, check the Local Setup Guide.

⚠️ This app was built as a university project submission for a course module.

🚀 Features

• 📷 Barcode-Scanner & Research: Snap a picture of a barcode or enter an EAN manually to immediately retrieve raw product details from a local OpenFoodFacts dataset.
• 🚦 Eco-Scoring & Analysis: Translates complex product data into a simple "Green-Score" (0-100) using a color-coded traffic light system. Features category ratings for environmental, social, and health impacts, complete with short AI-generated justifications.
• 📍 Local Alternatives & Geolocation: Recommends alternative products with a higher score available nearby. Uses geolocational data to show store distances and open directions directly in Google Maps.
• 📉 Impact Tracking & Scan History: Saves your scan history and calculates cumulative CO2 savings over time. Keeps users motivated through visual statistics and weekly push notifications.

⚙️ Architecture & Communication

EcoScan is structured as a distributed microservices system utilizing both synchronous and asynchronous communication:

graph TD
    App[React Native Mobile App] -- " REST (JSON) " --> Backend[Spring Boot Backend]
    Backend -- " JDBC " --> DuckDB[(DuckDB / Parquet)]
    Backend -- " JDBC " --> Postgres[(PostgreSQL)]
    Backend -- " 1. Publish Tasks " --> RabbitMQ{RabbitMQ}
    RabbitMQ -- " 2. Consume Tasks " --> AIService[crewAI Service]
    AIService -- " 3. Publish Results " --> RabbitMQ
    RabbitMQ -- " 4. Consume Results " --> Backend
    AIService -- " REST (Tool Queries) " --> Backend
    AIService -- " Local LLM Queries " --> Ollama[Ollama / Local LLM]
    App -- " OAuth2 Auth Flow " --> Keycloak[Keycloak]
    Backend -- " Token Verification " --> Keycloak

Loading

Communication Protocols

• REST APIs: Used for all synchronous operations. The React Native mobile app communicates directly with the Spring Boot backend to register scans, load history, and query alternatives. Additionally, the Python-based crewAI service performs REST requests to query databases or fetch specific tool-related information from the backend. Keycloak acts as the OAuth2 Identity Provider, securing all API endpoints.
• RabbitMQ Message Queue: Used for bidirectional asynchronous processing. For example, hen a product is scanned and needs an in-depth ecological analysis:
  1. The backend publishes a task to the ecoscan.ai.tasks.product-analysis queue.
  2. The Python-based AI service consumes the task and runs the crewAI agent pipeline.
  3. The AI service publishes the results to the ecoscan.ai.results.product-analysis queue.
  4. The backend consumes the results, updates the database, and pushes the final evaluation to the mobile app via Server-Sent Events (SSE).

Technology Stack

• Frontend: Expo, React Native (TypeScript), React Native Paper, React Native Reanimated, Expo Camera
• Backend: Java Spring Boot, Spring Data JPA, Flyway, Spring Security (OAuth2 Resource Server)
• Databases:
  • DuckDB: For high-performance analytical queries on the local Open Food Facts Parquet export
  • PostgreSQL: For storing transactional user data, scan history, and settings
• AI Service: crewAI (Python, uv, multi-agent orchestrator), Ollama (Local LLM runner)
• Identity & Access Management: Keycloak
• Message Broker: RabbitMQ

👥 Contributors