This repository contains our final work for the Computer Engineering graduation at the National Telecommunications Institute located in Santa Rita do Sapucaí, Brazil.
The identification of pests is a process with a high degree of complexity, as it requires specialized labor, considerable time and is subject to human failure. The implementation of a computer system to identify pests has helped farmers around the world, as it makes the process much faster, more efficient, and automatic. In this sense, the present work proposes the comparison of four models of Convolutional Neural Networks (CNN), through experiments with real images of the corn crop, and the collection of results with well-disseminated metrics.