Intelligent animal husbandry is becoming a priority area of the industry. On the basis of digital technologies, genomic assessment, and artificial intelligence, new opportunities are being formed to improve the organization of breeding and technological processes. For effective horse breeding, coupled with classical breeding methods, modern breeding resource management systems based on innovative approaches are needed. Accurate quantification of phenotypic information about an animal is a difficult task. Of particular importance there are the issues of objectification of animal characteristics by exterior due to the fact that the assessment of external forms is based on visual perception, is not devoid of a subjective approach and is subject to inaccuracies. One of the ways to solve this problem is to switch to a digital assessment of the phenotypes of interest. The article presents the results of the application of deep learning to solve the problem of automatic marking of characteristic points on a digital image of the studied objects. It was revealed that the created and trained neural network architecture as a whole demonstrated good accuracy.