Detecção Automática de Postes Elétricos e seus Componentes em Imagens de Nível de Rua via Métodos de Detecção de Objetos

Abstract

This article presents an approach to the inspection of the electrical network using the city of Campo Grande as a scenario. Advanced deep learning and computer vision techniques were used with the aid of the MMDetection library. The article performs a comparative analysis of several models in different versions of the dataset, evaluating the impact of data augmentation techniques on the robustness of the model. The results highlight the importance of these practices to genaralize the model, evidencing the efficacy of each technique implemented.