1. pygeovision-ghana-urban-analysis 2. landsat-change-detection-accra 3. pygeovision-showcase-accra 4. geoai-urban-expansion-ghana 5. accra-landsat-change-detection
# PyGeoVision: Accra Change Detection & LULC Analysis
## π Overview
This project demonstrates a complete geospatial analysis workflow using **PyGeoVision** - the unified platform combining satellite data access (22+ providers) with state-of-the-art GeoAI. The analysis focuses on **Accra, Ghana**, tracking urban expansion and vegetation change from 2020 to 2024 using Landsat 7 ETM+ imagery.
*Complete 9-panel analysis showing RGB composites, NDVI, change detection, and LULC classification*
## π Key Results
| Metric | Value | Visualization |
|--------|-------|---------------|
| **Urban Area (2024)** | 68.1% | ποΈ |
| **Forest Cover (2024)** | 0.0% | π³ |
| **Water Bodies (2024)** | 31.6% | π§ |
| **Net Vegetation Change** | +1.2% | π |
| **Buildings Detected** | 120 | π’ |
## πΈ Analysis Visualizations
### Land Use/Land Cover Classification (2024)
*Rule-based classification showing 5 land cover classes: Water (Blue), Urban (Brown), Sparse Vegetation (Light Green), Agriculture (Khaki), Dense Forest (Forest Green)*
### Vegetation Change Detection (2020 β 2024)
*5-class change detection: Severe Loss (Dark Red), Moderate Loss (Crimson), Stable (Gray), Moderate Gain (Lime Green), Severe Gain (Dark Green)*
### True Color RGB Comparison
| 2020 | 2024 |
|------|------|
| | |
*Landsat true color composites showing urban expansion and land cover changes*
### Building Footprint Detection
*Deep learning-based building extraction detected **120 building footprints** in the Greater Accra region*
### Spectral Indices Analysis
| NDVI 2020 | NDVI 2024 | NDWI 2024 |
|-----------|-----------|-----------|
| | | |
*Normalized Difference Vegetation Index (NDVI) and Water Index (NDWI) showing vegetation health and water bodies*
## π Features Demonstrated
### Data Acquisition
- **PyGeoFetch**: 22+ satellite providers (Planetary Computer, USGS, AWS)
- **Search & Download**: Automated scene selection based on cloud cover
- **Po β¦