Potholes are a prevalent problem that affects the effective functioning of the road infrastructure. They cause a high number of road accidents, and vehicle damage, resulting in higher maintenance costs. Effective strategies are therefore required to report potholes to road maintenance authorities and thus facilitate constant upkeep and repair. This study focuses on the development of a deep-learning-based pothole detection and reporting system, to assist road maintenance personnel in making informed road repair decisions. A YOLOv5 model is developed and trained using a custom dataset containing pothole images obtained from a survey conducted by the researcher, as well as online sources. The model is integrated into the overall system, to facilitate the detection of potholes from uploaded images. Comprehensive data visualizations are created, with the location of detected potholes added to a map. Pothole maintenance procedures are implemented to ensure that the pothole information in the system is accurate and up to date. Overall, the system provides a clear and concise viewpoint from which road repair decisions can be made and justified.