Aplicação web de um sistema de identificação e mapeamento de defeitos em pavimentos com utilização de inteligência artificial

Abstract

Due to the lack of financial resources and personnel in Pavement Management Systems and the need for periodic evaluations of Brazilian highways for more assertive decision-making regarding resource allocation, there is a need to develop low-cost tools to assist in evaluating the actual conditions of pavements and allow for the dynamic and interactive visualization of these data to improve understanding. In this context, this work aimed to develop a web application of a system for identifying and mapping pavement defects using artificial intelligence to assist decision-making by the Court of Auditors of the State of Ceará (CAS). To achieve this goal, a web application was created that used the georeferencing of pavement defects in five sections of Ceará's highways, which were obtained through video recordings of the sections and identified by means of an artificial intelligence model using convolutional neural networks. The application's results demonstrated that Google Maps was the most suitable API to meet the system's requirements, and that the combination of PostgreSQL with Strapi made the database more efficient due to its practicality and reliability. The Docker container approach proved to be suitable, as it has a tool capable of enabling the grouping, dependencies, and additional configurations of services in one place. To evaluate the application's efficiency, quantitative and qualitative analyses of the pavement conditions of the object section were carried out. Therefore, the use of a web application to visualize data linked to an artificial intelligence model to identify pavement defects to aid in decision-making proved to be quite effective.