Vehicle Detection And Accident Prediction In Sand/Dust Storms
Arun Singh, Dora Praveen Kumar, Kelothu Shivaprasad, Mohit Mohit, Ankita Wadhawan
Abstract
In this era of a smart and modern world that is designed by progressing technology, automated vehicles would become a precious part of it. The first thing that strikes in our minds talking about vehicles is traffic and accidents. Accidents could take place because of several reasons: dense traffic, unfavorable weather conditions, sudden braking, change in speed, etc, and the solution to this is machine learning, computer vision, and deep learning. Our focus is to improve the vision in areas of low visibility and predict the future by analyzing the present. Here we introduce a model which would help in dehazing and improving the visibility for a better driving experience in adverse weather especially targeting sandstorms and dust storms which would be quite common in the future because of the afforestation, the procedure is divided into two categories the dehazing and second is vehicle detection, situation analysis, and prediction. We have also incorporated things like estimating traffic density(dense/sparse), and the fire's in the worst situation using python, tensor flow, deep learning, and counting vehicles entering and departing from the frame.