Map to GeoJSON Converter

A free, open-source tool to convert map images (PNG, JPG) and SVG files into GeoJSON format, using AI and computer vision. Born as an alternative to expensive databases like Geochron (500€), enabling accessible geospatial data creation.

---

Table of Contents

• Highlights
• Background
• Quick Start
• Usage
• Features
• Algorithm Overview
• Outputs
• Limitations & Roadmap
• Contributing
• Documentation
• References

---

Highlights

• 🆓 Free & Open-Source: No costs, ideal for students and researchers.
• 🎨 AI-Powered: Uses segmentation and contour detection for automatic polygon extraction.
• 🗺️ Multiple Inputs: Supports images (PNG, JPG) and SVG files.
• 📦 GeoJSON Export: Outputs standard GeoJSON FeatureCollection.
• 🔬 Prototype: Extensible for research and education.

---

Background

This project was created by students to provide free access to geospatial data. Commercial services charge money for databases, making them inaccessible for educational projects. Our tool leverages open-source libraries (OpenCV, GDAL, PyTorch) to convert simple map images into usable GeoJSON files.

🎯 What It Does

Transform map images into GeoJSON automatically:

Input Map	Output (Detected Regions)

The tool extracts colored regions, identifies boundaries, and generates GeoJSON files ready for use in GIS applications.

---

Quick Start

Prerequisites

• Python 3.8+
• Libraries: OpenCV, NumPy, Shapely, PyTorch (optional for AI)

Installation

# Create virtual environment
python -m venv venv
venv\Scripts\activate  # Windows

# Install dependencies
pip install -r requirements.txt

Run

# Image to GeoJSON with AI
python "src/test con ai/image_to_geojson_auto.py"

# SVG to GeoJSON
python "src/test svg to geojson/Svg_to_Geojson_Converter.py"

---

Usage

1. Prepare your map image or SVG file.
2. Run the appropriate script.
3. Choose calibration (Italy preset or manual).
4. Outputs: GeoJSON file and debug images.

For detailed pipeline, see src/test con ai/pipeline.md.

---

Features

• Image Conversion: Extract polygons from map images using K-Means segmentation and contour detection.
• SVG Support: Convert SVG paths to GeoJSON.
• Georeferencing: Optional pixel-to-lat/lon calibration.
• AI Integration: Prototype with deep learning for better segmentation.
• Debug Visuals: Segmented and region overlay images.

---

Algorithm Overview

• Preprocessing: Image segmentation with K-Means or AI models.
• Contour Detection: Use OpenCV to find shapes.
• Filtering: Remove noise, water, etc., based on heuristics.
• Georeferencing: Map pixels to coordinates.
• Export: Generate GeoJSON with properties (id, color, area).

See src/test con ai/pipeline.md for full architecture.

---

Outputs

• GeoJSON File: FeatureCollection with polygons.
• Debug Images: _segmented.png (clusters), _regions.png (polygons).
• Visualize at https://geojson.io.

---

Limitations & Roadmap

• Works best on simple maps; complex ones may need manual tweaks.
• Linear calibration; advanced georeferencing planned.
• Future: Web interface, batch processing, better AI models.

See docs/feasibility/StudioDiFattibilità.md for detailed feasibility study.

---

Contributing

We welcome contributions! Open issues or PRs. Focus areas:

• Improve AI segmentation.
• Add more input formats.
• Enhance georeferencing.

For requirements gathering, see docs/feasibility/requirements/.

---

Documentation

• Feasibility Study: docs/feasibility/StudioDiFattibilità.md
• Requirements: docs/feasibility/requirements/Analisi_Concorrenza.md, Suggerimenti_Spontanei.md
• Pipeline: src/test con ai/pipeline.md
• Contributing: CONTRIBUTING.md
• Examples: examples/
• Tests: tests/
• Changelog: CHANGELOG.md

---

References

• Scripts: src/test con ai/, src/test svg to geojson/
• Inspired by open-source GIS tools like QGIS and GDAL.
• For feedback, open GitHub Discussions.

---