Weather Image Classification

The value of data has increased significantly throughout the years. It can be in the form of Text, Numbers, Images, Audio, or Videos. Images certainly make up a large amount of the available data. This application can be used for different purposes. This application, together with decision support systems for traffic communication, afforestation, and weather prediction, will be helpful in situations where weather forecasts are inaccurate, such as when self-driving cars must operate safely in dangerous weather. There are some previous works on this subject, but with fewer classes. This project focuses on eleven classes of weather images and to come up with the best model to predict the image as accurately as possible after training with many images. Since this application has importance in real-life situations, accuracy plays a vital role in avoiding harmful incidents. To accomplish the business goal, the model should be able to predict the rain even if it cannot determine the rain's speed from the image. This project can serve as the foundation for further weather prediction applications when a computer is required but human expertise can't be used and in situations where human expertise may be biased. In this project, it is assumed that weather may be classified in eleven different ways, from safe to dangerous, and it makes sense to divide it into these two categories, since the weather can be dangerous sometimes. This project aims to build a machine learning model to accurately classify the images into 11 classes namely dew, frost, glaze, rime, snow, hail, rain, lightning, rainbow, and sandstorm. In this project, the model will be trained on thousands of weather images to predict the weather condition in the Image. Finally, the objective is to develop a web Application which predicts the weather condition in the image uploaded by the user.