CLASSIFICATION OF THE TYPES OF THE COW’S UDDER TEATS USING A CONVOLUTIONAL NEURAL NETWORK

Abstract

The development of machine milking technologies requires the creation of methods for evaluating the parameters of udders and teats of dairy animals to assess their suitability for machine milking. Convolutional neural networks are used in the work as a tool for finding and establishing objects, including determining the boundaries of the nipples, and can be adapted to identify the type of structure of the udder nipples of animals. (Research purpose) The research purpose is developing a convolutional neural network that allows automatic assessment of animals by the type of structure of the teats. (Materials and methods) It was shown that the neural network used is based on the convolutional neural network VGG-16 - this is the architecture of a convolutional neural network for recognizing objects in static and dynamic scenes. The classification developed in FNAC VIM with the definition of six types of nipples was used to establish the type of nipples, the distribution of types is uneven. The training sample was adjusted to quickly select animals suitable for machine milking, while all three types of udder nipples considered in the study are suitable for it. (Results and discussion) We collected a data array of 110 images of udder nipples of animals of three selected types. A test sample of 40 images was used for image recognition, the average accuracy was 87.5 percent. It was found that the accuracy was 92 percent when recognizing images containing the first type of structure of the udder nipples, 78 percent for the second type of structure of the udder nipples and 83 percent for the sixth type of structure of the udder nipples. (Conclusions) It was revealed that the obtained results of the study can be used for the selection of nipple rubber and the adjustment of collectors in the milking parlor. It was stated that the correct selection of nipple rubber will exclude the occurrence of mastitis diseases from physical damage to the udder, will minimize the premature fall of the milking cups, which will not disrupt the milking process.