Automatic detection of lameness in cattle through 3D kinematic analysis and machine learning: database, training tool and locomotion score identification

• Grant number: 24/05468-5
• Support Opportunities: Scholarships abroad - Research Internship - Master's degree
• Start date: July 31, 2024
• End date: January 30, 2025
• Field of knowledge: Agronomical Sciences - Veterinary Medicine - Animal Clinics and Surgery
• Principal Investigator: Fabio Celidonio Pogliani
• Grantee: Victoria Portela Diniz Gaia
• Supervisor: Marina Von Keyserlingk
• Host Institution: Faculdade de Medicina Veterinária e Zootecnia (FMVZ). Universidade de São Paulo (USP). São Paulo , SP, Brazil
• Institution abroad: University of British Columbia, Vancouver (UBC), Canada
• Associated to the scholarship: 23/00338-3 - Automatic detection of lameness in cattle through 3D kinematic analysis and machine learning: database, training tool and locomotion score identification, BP.MS

Abstract

Lameness in cattle is one of the biggest health and welfare problems in intensive dairy production systems worldwide, in addition to generating economic losses to the producer with direct and indirect costs. The most used lameness evaluation and classification method is still by determining the locomotion score and, despite its low cost, it is a subjective technique that requires trained labor. An important limitation of this method would be that visually noticeable gait changes may be noticed later, after the animal has already had the injury for some time and, consequently, affecting the patient's productivity, health and welfare in a more prolonged way. Therefore, early and accurate identification of lameness is important so diagnosis and treatment can be carried out as soon as possible, preserving the cow's welfare and production and reducing the economic impact of this morbidity. In this way, automatic detection technologies have stood out because they are more objective, minimize disruption to the farm's routine, do not depend on labor and training people and enable early identification of changes in animals. Hence, the use of 3D cameras proved to be interesting for this purpose, as it reduces some disadvantages of 2D cameras and improves efficiency, along with being able to be used from a dorsal view, reducing the space needed on the farm. This research project proposes the identification of anatomical points in dorsal view, with the aim of correlating the movement dynamics of these structures in the animal's gait with the locomotion scores so that, later, a computational system capable of automatically diagnose and classify lameness in dairy cattle can be developed.