Uso de inteligência artificial no processo de detecção de manifestações patológicas de edificações - estudo de caso em edificação do patrimônio histórico: Igreja São Tomé (MG)

Abstract

The preservation of historical heritage is essential for maintaining a society’s cultural identity. However, historical buildings face significant conservation challenges due to aging, lack of maintenance, and limitations in traditional diagnostic methods. In this context, this monograph proposes the use of artificial intelligence (AI) as a complementary tool for detecting pathological manifestations in historical buildings. The case study was conducted on the Igreja Matriz de São Tomé, listed by IEPHA-MG and located in São Tomé das Letras (MG), applying non-destructive techniques such as visual inspection, conventional imaging, and infrared thermography. Subsequently, Microsoft Custom Vision®, a computer vision tool based on machine learning, was used to train models for the automatic recognition of pathological manifestations, with a focus on paint peeling. Three models were developed with different image datasets and evaluated using performance metrics such as precision, recall, and mAP. The results showed limitations in performance, but indicated potential for use as a complementary tool in building inspections. This work emphasizes the importance of integrated and multidisciplinary approaches, combining traditional methods and new technologies to optimize diagnostics, reduce subjectivity, and support the sustainable conservation of historical heritage.