An automated monitoring system for phenotypic characteristics of Reindeer (Rangifer tarandus) using AutoML technologies

Contemporary automated monitoring systems in animal husbandry and environmental protection rely on advanced computer vision methods to evaluate animal traits. This research introduces an automated animal condition monitoring system using the YOLOv11 convolutional neural network. The system was tested on specialized datasets featuring images of deer alongside their pre-measured attributes. Through the AutoGenNet software platform, processes such as hyperparameter tuning and architecture configuration were automated, streamlining and accelerating the adaptation of the model for monitoring various animal species. The findings highlight the effectiveness of YOLOv11 for this application. Additionally, the study validates AutoGenNet’s role in automating the development of models for monitoring reindeer phenotypic characteristics, supporting the integration of AI systems in modern animal husbandry. The biometric data obtained allow for the calculation of derivative metrics—like live weight, muscle mass, and reproductive potential—that inform management decisions. A significant accomplishment is the deployment of contactless monitoring, which removes stress linked to animal handling and adheres to bioethical standards. Successful trials on reindeer, which pose a complex biological challenge due to high variability in characteristics, ensure the method’s applicability to other livestock (such as pigs and sheep) under less stringent conditions. The developed automated monitoring system for reindeer phenotypic characteristics, based on YOLOv11 and AutoGenNet, has proven both technologically feasible and practically valuable.