Classificação de bovinos por meio da característica do focinho utilizando visão computacional

Abstract

The control of the cattle herd is a fundamental factor to add value to the product, both for dairy production and for beef cattle. The one that makes the most use of this method is the beef cattle, due to the requirements of importing countries. For this reason, a better identification is necessary because the methods currently available, such as radio frequency identification (RFID), earrings, red-hot branding, have problems because they are subject to fraud and they cause suffering to the animal. In order to solve the issues mentioned above, this work proposes to use classification methods for each of the animals and in several environments that are necessary. In this way, this work uses computer vision to classify cattle by muzzle characteristics. Techniques with gray level co-occurrence matrix, genetic algorithms and knn classifier are used. The results of the study demonstrate that the hit rate of this method is 96.75%, thus showing that the classifier can be used for the proposed purpose.