Detecção automática de aves em imagens coletadas com aeronaves remotamente pilotadas para uso em gerenciamento de risco à vida selvagem em aeroportos

Abstract

Wildlife strikes to civil aircraft frequently occur worldwide and are a major safety problem. In the United States alone, from 1990 to 2019, approximately 209,950 strikes were recorded, causing material damage and loss of life. In this study, we aimed to develop a method for using images acquired by remotely piloted aircraft and a process of automatic detection of patterns in such images to create a robust monitoring system for detecting animals and birds at the Coe Field private airport (8FA4), with a hit rate of 96% or higher for the animals and birds present in images at the time of flight.