Body Condition Score (BCS) Classification with Deep Learning

Abstract

The most important indicator of whether animals 'needs are met in livestock enterprises is the animals' body condition score (BCS) score. In dairy cattle BCS is based on scoring from 1 to 5 according to the external appearance of the animals. BCS is a subjective method based on visual or palpation method to determine the relationship between subcutaneous fat thickness and bone protrusions in pelvic region in back, waist and coccyx regions in cattle. Generally, BCS values in the enterprises are determined by a method based on expert knowledge and determined by observation. If the animal is above or below the desired BCS, at this stage, diseases resulting from metabolic problems, low yield or animal losses may be observed. With the regular control of this situation, the profitability of the enterprise may increase with the production of more health animals. For this purpose, it was aimed to determine the BCS score with a computer-aided software. Images from cattle were arranged in specific forms and classified by Convolutional Neural Networks (CNN). Of the 180 images, 75% were used for training and 25% for testing. In this study, system performance was increased by using pre-trained CNN architectures and the responses of different architectures to BCS classification problem were tested. As a result, it was seen that BCS scoring can be done more than 60% successfully by using CNN methods.