An automated classification for danger of extinction Animals from Colombia using convolutional neural networks

Abstract

The extinction of different types of animals is a problem that has been growing over the years, and that, consequently, has caused environmental problems, such as climate change. Genetic diversity (biodiversity) is essential for the development of all species and human beings depend on it in their daily lives. When biodiversity decreases, human life expectancy is reduced, not only from an ecological point of view, but also from a resource point of view, even to be able to have species that are adapted to an ecological niche. This research will expose a computer strategy that over time has achieved great results; convolutional neural networks is a process that has facilitated the monitoring of different kinds of animals in recent years, this, in order to facilitate the process of recognition and counting of animals, focused on agriculture and zoology. For this, an architecture in the field of convolutional neural networks (CNN) will be used, Alexnet, which has references with very high results. In addition, the mathematical programming software Matlab is used for the development of the neural network. Getting of this way a result of accuracy of validation of 97,52%, with the use of a dataset with 3026 images, in where, 80% are used for training and 20% for validation.