Above-Ground Biomass Prediction
Upload a multi-band satellite image to predict above-ground biomass (AGB) across the landscape.
Requirements:
- Image must be a GeoTIFF with spectral bands
- For best results, use imagery with at least 59 bands or similar to training data
About This Model
This biomass prediction model uses the StableResNet architecture to predict above-ground biomass from satellite imagery.
Model Details
- Architecture: StableResNet
- Input: Multi-spectral satellite imagery
- Output: Above-ground biomass (Mg/ha)
- Creator: vertify.earth for GIZ Forest Forward
- Date: 2025-05-19
How It Works
- The model extracts features from each pixel in the satellite image
- These features include spectral bands, vegetation indices, texture metrics, and more
- The model outputs a biomass prediction for each pixel
- Results are visualized as a heatmap or RGB overlay
Updates in This Version
- Fixed biomass value calculation issue (improved log transform handling)
- Added detailed diagnostics for troubleshooting
- Enhanced RGB visualization with band verification