Object detection in order to determine locations for wildlife crossings

Abstract

The intensive construction of road infrastructure due to urbanization and industrialization around the world carries with it negative environmental impacts, primarily due to increased emissions of gases, but also due to the separation of natural habitats and ecosystems. In order to overcome this problem, without affecting the mobility of the population, it is necessary to allow wild animals to cross over or below the roads, i.e. to create wildlife crossings, which requires knowledge of the locations where the corridors of animal movements intersect with existing or planned roads. This paper analysis the establishment of a camera system and the application of a deep learning methodology for the automatic identification of animals by species and number, in order to determine locations for the construction of crossings for large wildlife. Also, the paper presents the possibility of using geographic information systems to analyze information obtained by monitoring built wildlife crossings.