An Intelligent Pothole Detection and Alerting System using Mobile Sensors and Deep Learning

One of the increasing problems commuters face on a day-to-day basis are terrible road conditions, specifically potholes. There is a disconnect between people, elected representatives and civil agencies in charge of fixing potholes. The problem lies in both, detecting the potholes and the fact that concerned departments are not able to coordinate to resolve the issue. The communication gap is bridged using two mobile applications, one for the public and one for the government, with distinct features and latest technology. The mobile applications are built using Flutter, a cross platform for Android and iOS. The public app is used by the commuter to report potholes through various modes such as travel mode and capture image mode. These modes use inbuilt smartphone accelerometers and deep learning techniques such as YOLOv3 models. They are also provided with special features like alert mode that is used to alert the user of nearby potholes while commuting as well as geotagging of all potholes detected by all users. The tailored mobile application for Elected Representatives, PWD and the assigned contractor are built to monitor the reporting and progress of fixing potholes and to assign the next task in the process.