Capturing urban green view with mobile crowd sensing

Abstract

Urban green spaces are beneficial to ecosystems and the health of people. The Green View Index (GVI) is an essential metric for assessing urban green spaces from a human perspective. However, measuring GVI at an urban scale requires extensive collection and processing of sensing data, posing challenges in terms of high resource consumption, difficulty in implementation, and lack of user participation. Mobile crowd Sensing (MCS) is an emerging large-scale, low-cost solution for sensing data collection. To address the aforementioned issues, this study proposes an MCS system called GreenCam to measure the GVI with smartphone sensors. GreenCam guides users to capture photos of urban green spaces from human perspective. The system employs a Transformer-based model, which is trained on a customized dataset of 1200 carefully-labeled urban green images, to extract the greenery from the captured photos and calculate GVI. With widespread participation from urban users, the photos captured by users with GreenCam can cover various streets and areas of the city, enabling the measurement of GVI at an urban scale. Additionally, these photos reflect people's preferences towards specific urban landscapes, and analyzing the distribution and characteristics of popular landscapes contributes to the enhancement of urban ecosystems and landscapes.