Pothole Detection Using Machine Learning Algorithms

Abstract

Potholes are holes on the surface of roads that are horizontally more than 75 mm and by depth more than 20 mm which are created naturally from weak construction, stuck water in rainy season, decay of rocks and also by overload vehicles and sometimes by all of them altogether. Statistics show that in our country in the last 20 years more than 57,000 people lost their lives on roads and a lot more were injured in 58,208 accidents. And a major percentage of these accidents are caused by potholes. The bikers nowadays are in grave danger because of this pothole. A good detection system is necessary to detect the potholes real time and alert the drivers to avoid any inconvenience or casualty. There is some work done on this specific problem, but we propose an approach to detect potholes real time by machine learning which will fulfil the requirements. We have applied some machine learning algorithms on our collected dataset’s features and got promising results. And our model will help save a lot of lives by detecting potholes real time.