An Intelligent Pothole Management System Using Multi-Tasking CNN on Android

Potholes on the road can pose significant risks to drivers and pedestrians. They are a common cause of accidents and vehicle damage. Hence, their timely repair is critical for road safety. However, pothole repairs are frequently delayed due to lack of information about their location and risk level. To address this issue, we propose an Intelligent Pothole Management System (IPMS) that promotes using citizen-sourced information through an android application in a citizen-government collaborative environment. The android application employs a novel lightweight multitasking CNN to detect and classify potholes based on their risk level instead of traditional computer vision methods that rely on calculation of geometric dimensions. The proposed model runs on the smartphone, compared to a server environment, as it relies on perspective vision instead of exact calculations and is trained using a carefully curated dataset of about 30,000 images. Our system, similar to Google Maps’ traffic view, shows citizens the pothole density as well as pothole severity on each road to avoid potential accidents. Citizens report potholes through the application which captures geo-location and extracts pothole information (risk level). Similarly, public works department as access to pothole relevant data in a dashboard, which they can use to prioritize repairs based on the risk levels of potholes and their location. By allowing citizens to report potholes directly, we hope to foster a sense of community ownership and responsibility for road safety. Our proposed system can also assist public works department in making more informed repair decisions, resulting in more efficient repairs and safer roads. We believe that our Intelligent Pothole Management System (IPMS) has the potential to improve road safety and is a valuable contribution to the development of intelligent transportation systems.