Aprendizagem de máquina aplicada ao monitoramento da presença de animais em reservas naturais de ambientes industriais.

Abstract

Several Brazilian industries consume natural resources on a daily basis. Some of them have their own regions of environmental reserves. In this context, these industries continually aim to provide protection for their resources. However, a complicating factor present in most of these spaces is animals. Your habits sometimes lead to damage to these areas. In this sense, this article aims to monitor the presence of animals in natural reserve environments in industrial environments. For this purpose, digital image processing techniques were used to identify the presence of animals in these environments. Due to the momentary impossibility of acquiring images in an industrial environment, the work presents a proof of concept, with the acquisition of data from a public database. Deep learning networks were used to identify animal images. The presented approach enabled the identification of animals of different sizes and species, with relevant accuracy in comparison with other similar studies. These results allow to validate the proposed methodology, enabling itsinstallation in an industrial environment, in order to monitor the presence of animals.