Real-Time Green View Index Assessment: AI-Driven Analysis of Urban Vegetation Using Dashcam Videos

The Green View Index (GVI) is a valuable metric for assessing visible greenery in urban areas and improving urban planning. However, current GVI calculation methods, which predominantly rely on manual assessments through on-site visits or Google Street View images, often provide outdated data and fail to account for the quality and health of vegetation. This paper proposes an AI-driven approach to improve the accuracy and timeliness of GVI measurements using dashcam videos for real-time vegetation analysis. A deep learning model based on YOLOv11 is used for semantic segmentation of vegetation and terrain. The GVI calculation is further enhanced by incorporating vegetation health indicators derived from pixel saturation and hue values. Greenery is categorized into two distinct classes: “vegetation,” which represents vertical greenery (e.g., trees), and “terrain,” which represents horizontal greenery (e.g., grass). This classification allows us to prioritize vertical greenery, by giving it greater weighting, which generally offers greater ecological and aesthetic benefits in urban environments. In addition, GVI heat maps are generated to visually represent the distribution of greenery across Doha, Qatar. This approach overcomes the critical limitations of static GVI models, offering a more dynamic and accurate tool for assessing urban green spaces by enabling real-time, health monitoring, and vertically-prioritized greenery assessments. This allows for making data-driven decisions when addressing sustainable urban planning. Adaptive policy making, improved ecological designs, and consistent audits for urban greenery could all become possible due to this proposed method. This is particularly valuable in rapidly growing regions like Qatar and regions with limited street-view coverage, where timely vegetation data is essential for sustainable urban planning.