Application of Artificial Neural Networks in the Classification of Pathological Manifestations Using Images Captured by a Remotely Piloted Aircraft

Abstract

With today's technological advances, many areas have made great leaps forward by incorporating and incorporating automation systems into their routines, and the broad field of modern architecture and civil construction have followed the same path. Programming is becoming a highlight around the world and with it we know artificial neural networks, sets of algorithms that, when trained, can identify and classify different elements. The objective is to use Remotely Piloted Aircraft (ARP), extract images to perform image processing using software and train neural networks in order to diagnose pathological manifestations in modern building architecture. In turn, pathological manifestations, as they are post-construction elements, are identified and classified by civil engineering professionals, who have increasingly adopted the use of technological tools to facilitate their routines. With the use of images taken through cameras and drones to identify pathological manifestations, neural networks began to play a fundamental role when it comes to identification, classification of these elements and even efficiency in activities. The use of the elaborate neural network served to identify where these pathological manifestations were located and what they were, with only real images of the damaged building and thus could be compared with a damage map, finally having an acceptable accuracy in relation to its results.