The use of machine vision technologies to monitor the behaviour of dairy cattle

Publication date: 25-03-2026
Extent: 40 pages

Contributions by: Oleksiy Guzhva, Swedish University of Agricultural Sciences, Sweden; Marwa Mahmoud and Ozgur Civan Dogan, University of Glasgow, UK; David Berthet, Sony Nordic (Sweden), Sweden; and Niclas Högberg, Swedish University of Agricultural Sciences, Sweden

Chapter synopsis:

The dairy industry faces increasing pressure to enhance productivity, welfare, and sustainability. Precision Livestock Farming, integrating advanced technologies, has emerged as a key strategy. Within this framework, computer vision (CV) systems enable continuous, non-invasive, automated monitoring of dairy cattle, improving traditional methods reliant on manual observation or wearable sensors. Recent machine learning and deep learning advances have enhanced CV-based detection of diverse behaviours and health indicators, including lameness, body condition, feeding patterns, and estrus. Emerging techniques, such as pose estimation and facial expression analysis, further deepen understanding of subtle behavioral changes. However, challenges persist. Limited standardized datasets hamper algorithm development and comparison, while differences in perspective between animal and computer scientists complicate interdisciplinary collaboration. Variability in farm environments and ethical considerations also limit widespread implementation. This chapter reviews current CV applications, methodologies, and limitations, providing guidance for experiment planning and integration. It ultimately advances both animal science and computational innovation.

DOI: 10.19103/AS.2025.0167.05

Click here to download