24th EANN 2023, 14 - 17 June 2023, León, Spain

Semi-automated System for Pothole Detection and Location

Favell Nuñez, Rodrigo Espinal, Gerardo Zavala, Josue Banegas


  Imperfections in land transport infrastructure can be made up of various elements. As it will be seen in this report, these imperfections vary depending on the variables that make up their formation. However, one of the most damaging elements to cars is made up of large, deep blemishes also known as potholes, which present a constant threat to the integrity of cars. One of the problems with the process of recognizing potholes is that, in countries like Honduras, these routes are made on foot, which implies a great amount of time, effort and resources for those in charge of recognizing potholes. To solve this problem, a solution was proposed that would speed up the recognition process. This solution consists of an artificial intelligence model that was trained to recognize potholes, using a database made up of pothole images taken from the internet and pothole images captured in Tegucigalpa, which resulted in a model with a maximum precision value of 86.251% and a maximum sensitivity value of 80.035%. Along with this model, a VK-162 GPS module was used, which oversaw extracting the geographical location of each pothole after being detected. With these geographical points it was possible to map the coordinates corresponding to each pothole on a map to represent a pothole recognition route more easily and efficiently and thus creating a system that could identify and locating potholes in a semi-automated manner.  

*** Title, author list and abstract as seen in the Camera-Ready version of the paper that was provided to Conference Committee. Small changes that may have occurred during processing by Springer may not appear in this window.