GitHub

Introduction to GeoRetina AI (GRAI)

Welcome to GeoRetina AI (GRAI), a powerful platform delivering a conversational, ChatGPT-like experience tailored for advanced geospatial analysis. GRAI's mission is to democratize geospatial insights at scale by harnessing cutting-edge AI technologies, making complex analysis accessible through natural language.

What is GeoRetina AI (GRAI)?

GRAI bridges the gap between complex geospatial data and actionable insights. It allows users to interact with vast datasets using simple prompts, eliminating the need for extensive coding or specialized software knowledge. With GRAI, you can:

  • Query and analyze diverse geospatial data (both satellite imagery and vector features) using natural language.
  • Visualize results dynamically on interactive maps.
  • Perform complex remote sensing and vector operations seamlessly.
  • Integrate custom knowledge bases (using Retrieval-Augmented Generation - RAG) to combine geospatial findings with your specific domain documents.
  • Upload and utilize your own vector data for tailored analysis.

Key Capabilities

GRAI offers a growing suite of analytical tools accessible via its intuitive chat interface:

  • Remote Sensing Analysis (Raster):
    • Urban Heat Island (UHI) Analysis: Evaluate temperature variations in urban landscapes.
    • Land Use/Land Cover (LULC) Mapping: Classify land cover types using models like Google DynamicWorld or custom AI models.
    • LULC Change Detection: Identify and quantify changes in land use over specified periods.
    • Air Pollution Monitoring: Analyze patterns and trends in air quality indicators (under development).
  • Vector Data Analysis:
    • Feature Querying: Ask natural language questions about vector datasets (e.g., "Show me all parks within 5km of this point," "Count the number of buildings in this area").
    • Data Integration: Combine insights from vector data (like infrastructure locations) with remote sensing analysis (like vegetation indices).
    • User Data Integration: Upload your own points, lines, or polygons (e.g., Shapefiles, GeoJSON) and analyze them alongside other datasets.
  • AI & Knowledge Integration:
    • Conversational AI: Interact naturally with the AI assistant to guide analysis and interpret results.
    • Retrieval-Augmented Generation (RAG): Upload documents (PDFs, etc.) to create a knowledge base. The AI can then synthesize information from your documents and the geospatial analysis results.

Core Technologies

GRAI leverages a robust, cloud-native architecture:

  • Google Earth Engine (GEE): For large-scale remote sensing data access and processing.
  • Vector Databases & Spatial Libraries: For efficient querying and analysis of vector data (e.g., OpenStreetMap, Overture, user data).
  • Advanced AI Models: Including Large Language Models (LLMs) for natural language understanding and custom Vision Models (via Vertex AI) for specific tasks.
  • Cloud Infrastructure: Built on reliable platforms like Google Cloud Platform (GCP) and Vercel.

Open Source Foundation: Chat2Geo

GRAI is built upon the foundation of our open-source project, Chat2Geo. While GRAI offers a managed, feature-rich, and commercially supported platform, Chat2Geo (available on GitHub) provides the core engine for community contributions and development. This documentation focuses exclusively on the capabilities and usage of the GeoRetina AI (GRAI) platform.

Explore GRAI

Dive deeper into what GRAI can do for you: