England's ambitious Environmental Improvement Plan aims to ensure every citizen lives within a 15-minute walk of a high-quality greenspace. Achieving this requires precise, frequent monitoring of urban and rural tree canopies, a task that has historically relied on expensive LiDAR (laser-based distance measurement) and labor-intensive field surveys. Forest Research, the UK's forestry agency, is now revolutionizing this process by integrating Meta’s DINOv2.

This open-source computer vision model was trained on 18 million satellite images to create a global canopy height map. By applying this technology to national aerial photography, researchers can now detect individual trees and estimate timber volume loss with unprecedented 1-meter resolution. This shift from costly specialized hardware to AI-driven analysis allows for a rolling three-year update cycle, ensuring policy decisions are backed by the most current data available.

The collaboration highlights the growing utility of Foundation Models in environmental science. While traditional methods struggle to identify small woodlands or lone urban trees, DINOv2 excels at identifying complex structures in diverse environments. As Meta introduces DINOv3, the UK government is poised to further enhance its visual intelligence capabilities, setting a global precedent for using open-source AI to achieve national sustainability and public health goals.

England's ambitious Environmental Improvement Plan aims to ensure every citizen lives within a 15-minute walk of a high-quality greenspace. Achieving this requires precise, frequent monitoring of urban and rural tree canopies, a task that has historically relied on expensive LiDAR (laser-based distance measurement) and labor-intensive field surveys. Forest Research, the UK's forestry agency, is now revolutionizing this process by integrating Meta’s DINOv2.

This open-source computer vision model was trained on 18 million satellite images to create a global canopy height map. By applying this technology to national aerial photography, researchers can now detect individual trees and estimate timber volume loss with unprecedented 1-meter resolution. This shift from costly specialized hardware to AI-driven analysis allows for a rolling three-year update cycle, ensuring policy decisions are backed by the most current data available.

The collaboration highlights the growing utility of Foundation Models in environmental science. While traditional methods struggle to identify small woodlands or lone urban trees, DINOv2 excels at identifying complex structures in diverse environments. As Meta introduces DINOv3, the UK government is poised to further enhance its visual intelligence capabilities, setting a global precedent for using open-source AI to achieve national sustainability and public health goals.