SYSTEMATIZATION OF METHODS OF IIDENTIFICATION OF COWS

Currently, there are a significant number of scientific publications on remote identification of cows, while there is a shortage of systematizing and generalizing scientific methods. (Research Purpose) The research purpose is systematizing existing methods of identification of cows. (Materials and methods) A search of scientific literature was conducted on the databases Elibrary, Scopus, IEEE Xplore, Google Scholar; as well as MDPI. The review and systematization of methods of identification of cows developed over the past 5 years are presented, as well as an analysis of scientific papers involved in issues on the topic of identification of cows, the main trends in this area are identified. (Results and discussion) We have found that the most popular identification methods are deep learning neural networks such as YOLO and CNN, which are used in 70 percent of cases. At the same time, identification of cows by the pattern on the body is used more intensively, and not by ear tags. On average, the declared accuracy of identification is 95 percent. However, some methods demonstrate higher accuracy, exceeding 98 percent. (Conclusions) It has been established that the further development of methods for identifying cows will be due to the use of deep learning neural networks using biometric features, such as a drawing on the body. It was found that the identification method based on the drawing on the cow's body has a number of disadvantages, namely: dependence on the pattern of wool, which can be deformed or contaminated, which causes problems with identification. And since systems are often tied to a specific breed on a specific farm, then to use the system on other farms and on other breeds of animals, it is necessary to re-train the neural network. A promising development of the direction for compiling digital three-dimensional models of cows has been identified, which will allow observations to be carried out using the same method for different breeds of animals.