In this project work, the main motive is to build a deep learning model to detect air pollution from real-time images. In order to achieve that goal, we have collected data from different sources and then enhanced the low-quality images using the Image enhancement technique. Our next step was to train a CNN (Convolutional Neural Network) on the images in order to detect air pollution by analyzing the clearness of the sky in the image. In this work, we have used the Inception V3 model. After the successful testing of the CNN model, we have deployed the model on an Android Application.