It is no doubt that air pollution influences human health. The report of diseases related to and numbers of patients suffered from air pollution increase rapidly over time. Hence, the requirement of measuring the air quality index (AQI) precisely and economically becomes the utmost purpose of communities. Although the most precise AQI comes from high-end stations, there is a problem with deploying such stations to cover all corners of particular areas. Some replacement methods are used to measure AQI using other data sources than from stations such as satellite, UAV, google street views, SNS, and open data from the Internet. This paper introduces a method that can predict AQI at a local and individual scale with a few images captured from smartphones and open AQI and weather datasets by utilizing lifelog data and urban nature similarity. Image retrieval and prediction model approaches are developed and evaluated on different open datasets of air pollution, weather, and images. The results confirm our hypothesis about the high correlation between the AQI and the surrounding environment’s snapshots.