Abstract
This paper presents a novel method for estimating the height of urban trees using UAV imagery and contextaware algorithm that leverages Digital Surface Models (DSM) and street-level data. Building upon previous work on automatic tree detection from drone imagery using machine learning, we developed an algorithm to estimate the approximate height of each urban tree. Our approach uses the assumption that street levels represent local ground reference and calculates tree height by measuring elevation differences between tree canopy areas and their nearest street segments. These data represent a valuable asset for those responsible for public vegetation in urban districts. Experiments conducted over a $2.95 \text{km}^{2}$ area demonstrate that the proposed method provides consistent and accurate height estimates with minimal overhead. The results highlight the feasibility of applying this method to urban forest inventories and support its use a complementary tool for city management and planning.