Effectiveness and producers perceptions of camera-based technology detecting hoof lesions In dairy cows

Abstract

Lameness remains a significant concern in the dairy industry, with growing interest in automated technologies for early detection and intervention. This dissertation combines quantitative, qualitative, and machine learning approaches to assess both the technical performance of an autonomous camera system and stakeholder perspectives on technology implementation. The first objective of this thesis was to evaluate how previous research has applied machine learning methods for detecting lameness and hoof lesions in dairy cattle (Chapter 1). The second and third objectives assessed the performance of an automated camera-based system that scores locomotion and body condition. Chapter 2 examined whether the system's locomotion scores were associated with hoof lesion outcomes, using hoof trimming data to compare cows with and without lesions. Chapter 3 evaluated the system’s ability to reliably identify individual cows. Chapter 4 assessed inter- and intra-observer reliability across different methods of body condition scoring, including human observers, photo-based scoring, and the automated system. Recognizing the role of human perspectives in technology adoption, Chapters 5 and 6 explored perceptions of lameness and lameness detection technologies among dairy farm decision-makers. These chapters focused on stakeholders view lameness management priorities and barriers to adopting automated technologies. Finally, Chapter 7 evaluated existing locomotion score-based thresholds for identifying cows with hoof lesions, while Chapter 8 developed a machine learning algorithm to improve classification of cows requiring intervention for hoof lesions. Together, these chapters contribute to advancing both the technical capabilities and real-world applicability of autonomous lameness detection technologies in the dairy industry.