Data augmentation approach in detecting roof pathologies with UASs images

Abstract Machine learning and computer vision techniques contribute to the automation roof pathologies identification from images collected with Unmanned Aerial System (UASs). However, one of challenges for practical machine model tuning is small-data problem. One strategy adopt data augmentation generating more training existing images. This paper evaluates in detecting inspections UASs The study adopted two models an image processing system. tests using obtained superior results accuracy, precision, recall, F-score, negative specificity metrics compared only original photos. These indicate that improves system’s performance identifying UAS inspection system proposed such integrated technologies would make it possible increase transparency, simplify steps reduce time perform inspections, streamlining preparation reports application corrective actions.