Estimating Individual Tree Height and Species from UAV Imagery

A team of researchers has developed a new, efficient AI system for large-scale forest monitoring using only standard drone cameras. Their model, DINOvTree, is a unified approach that uses a Vision Foundation Model (VFM) as a backbone with specialized heads for two critical tasks: estimating the height of individual trees and classifying their species. This dual-purpose design is trained and evaluated on a new benchmark dataset called BIRCH-Trees, which spans three distinct forest biomes: temperate, tropical, and boreal plantations. The goal is to provide a cost-effective and scalable alternative to traditional, labor-intensive field surveys for measuring forest biomass, a major global carbon sink.

In extensive evaluations, DINOvTree outperformed other common computer vision methods and biological allometric equations. A key finding is its remarkable parameter efficiency; it delivers accurate height predictions and competitive species classification accuracy while using only 54% to 58% of the parameters required by the second-best AI approach. This efficiency is crucial for deploying the model on edge devices or processing vast datasets from drone surveys. The work establishes a new standard benchmark (BIRCH-Trees) for the field and demonstrates how foundation models can be adapted for precise, real-world environmental science applications beyond general object recognition.