Evaluating the contribution of city-wide street trees on PM2.5 removal based on Street View Imagery and computer vision technology

Rapid urbanization and industrialization have resulted in significant environmental challenges in many cities worldwide, with pollution of particulate matter with a diameter of 2.5 μm or smaller, notably PM2.5, being a major concern. Street trees, as an integral component of urban ecosystems, hold the potential to mitigate this issue through their capacity to absorb and disperse PM2.5. However, the current assessment of street trees' ability to remove PM2.5 is prohibitively costly for large-scale implementation. This study introduces and evaluates a novel computer vision methodology that automates street tree profiling using Street View Imagery, focusing on characteristics such as tree height, crown and trunk diameters, and species. We applied our approach to measure PM2.5 removal capacities of street trees in the urban area of Jinan, China. In total, 98,009 street trees were identified, with over 90% belonging to eight dominant species. By utilizing this detailed tree information and in conjunction with real meteorological data, we simulated the PM2.5 removal process by street trees, found that street trees can annually eliminate approximately 1.6 tons of PM2.5 in the 258.27 km² study area with average 1.63 g per tree. The results underscore the profound capacity of street trees to alleviate the issue air pollution. This proposed research method is critical for informed urban planning and targeted environmental initiatives acorss global cities especially those in global south. © 2026 Elsevier Ltd