Roadside Tree Image-Based Air Quality Estimation of Gurugram City, Haryana, India

The wellness of people is at risk due to air pollution, hence accessible and efficient air quality monitoring is required. Even though traditional ground station-based monitoring systems are accurate, they frequently lack wide spatial coverage and are costly. Recent developments in AI and computer vision have made image-based air quality measurement a viable substitute. In order to evaluate outdoor air quality levels, this study suggests a novel method that makes use of roadside tree photos. Through the use of visual indicators from atmospheric conditions and leaf deterioration, we present a novel, non-invasive approach to air pollution measurement. We also introduce the Image Air Quality Index Database (IAQID), a comprehensive dataset gathered over a year to support further studies on image-based air quality estimation. To solve this issue, and we mostly gather excellent outdoor photos of trees on the roadside in high-medium-low traffic areas. We anticipate this database will stimulate more investigation into image-based air quality assessment. Furthermore, we use this database to conduct extensive studies. We examine the variations in the appearance of outdoor photos under various air quality circumstances using various handcrafted features. The study's findings indicate that weather characteristics are far more accurate than traditional handcrafted ones. In addition, the extinction coefficient, which has an accuracy of 68, quantifies the quantity of light that particles attenuate and functions best in environmental conditions.