Air pollution has continued to negatively impact humans and the planet in various ways, and air quality monitoring systems are essential for a deepened understanding of air pollution and its influence. In my project, air quality was analyzed by estimating air pollution from various natural webcam images, focusing on levels of particulate matter (PM2.5), a common pollutant. I developed a machine learning model (ResNet9) that can classify these webcam images into their corresponding levels of PM2.5 concentrations. Classification was accomplished by extracting certain features from these images, specifically haziness/visibility. To perform this classification, a combined dataset was created with webcam images from Beijing and Yosemite National Park, and their corresponding levels of PM2.5. After training and testing the model, the results showed optimal performance of the ResNet9 model, therefore validating this method of PM2.5 concentration estimation. The developed ResNet9 monitoring system can decrease costs and increase the implementation of monitoring systems, which will improve the human condition by motivating people to engage in reducing air pollution and push governments to establish air quality monitors.