Image Analysis Based on Var-Lstm Method for Air Quality Prediction

Abstract

Abstract

Data analysis and computer vision are two strong technologies that may aid us in accomplishing activities that would otherwise need more time and effort. Therefore, the first step in saving the environment should be to check air quality, particularly in developing nations. Current methods for gauging air quality need expensive, specialized instruments and infrastructure. Therefore, it might be challenging, if not impossible, to provide air quality data for far-flung locations or attractive regions, even inside cities, owing to the high price or technical complexity of the equipment required. Based on an investigation of several publicly accessible photographs of the surroundings of Beijing, Shanghai (China), and Phoenix, the authors of this paper offer a method for quantifying PM air pollution (US). The photos were processed to extract six elements that were then utilized in conjunction with other variables such as the time of day, location, and weather to forecast the PM^2.5 index. This was achieved by using deep learning techniques, namely the training of a VAR-LSTM model on the aforementioned picture dataset. The results show that PM prediction is feasible using the image analysis technique._2.5.