Canopy Height Maps v2
4 days ago
- #AI-models
- #forest-mapping
- #climate-action
- Meta and the World Resources Institute announce Canopy Height Maps v2 (CHMv2), an open-source model for detailed forest mapping.
- CHMv2 uses DINOv3, Meta's self-supervised vision model, for improved accuracy and detail in measuring tree canopy height globally.
- The model's R² score improved from 0.53 to 0.86, enhancing reliability for scientific and operational use.
- CHMv2 aids in monitoring forest health, restoration efforts, and carbon storage estimation.
- Collaborations with governments in Europe and the U.S. utilize CHMv2 for climate migration, biodiversity, and urban planning initiatives.
- Future improvements aim to address data-sparse regions and extend temporal coverage for better change detection.