Published Paper Details:

The health effects of air pollution endanger human lives, especially among the at-risk population and those who suffer from respiratory illnesses. Hence the detection and monitoring of air pollution becomes necessarily important. Given the exorbitant cost of putting highly precise air pollution monitors throughout a city is quiet expensive.. In contrast to previous machine learning studies that primarily focus on pollutant estimation based on single day-time images, our proposed deep learning model integrates ResNet with Long Short-Term Memory , extracting spatial-temporal features of sequential images taken from smartphones instead for estimating air pollution using high equipped sensors and other sophisticated euipments.. In this study, logistic regression is used to determine whether or not a data sample is polluted. Based on prior PM2.5 data, auto regression and other extracted features, ResNet is used to predict PM2.5 levels. Knowing the level of PM2.5 in the coming years, months, or weeks allows us to reduce it to a safe level. This proposed work aims to determine air quality level based on data set of camera images taken from a specific city in various time intervals.