Air-pollution prediction in smart city using convolutional long short-term memory model

People all over the world could die from the air we breathe. We cannot do much to stop air pollution until we know more about it and guess what will happen. We use the cutting-edge graph convolutional network (GCN) design and the cutting-edge deep prediction convolutional long short-term memory (ConvLSTM) model to guess what the PM2.5 levels will be in different parts of the city every hour over time. We cannot get data for our deep learning model from studying particles or the air. We use info from many places instead, like camera data and pictures. The TROPOMI multispectral imager on Sentinel-5P works as it should. The name of it is 'fire radiative power'. This is true because the PM2.5 data we used to train our deep-learning model is typical of the data from the ground. We can get a better picture of the PM2.5 level.