Reticulo-ruminal motility is a well-established indicator of gastrointestinal health in dairy cows. The currently available methods for assessing motility are labor-intensive, costly, and impractical to use regularly for all cows on a farm. We hypothesized that the reticulo-ruminal motility of dairy cows could be assessed automatically and remotely using a low-cost 3-dimensional (3D) camera. In this study, a 3D vision system was constructed and mounted on the frame of an automatic milking robot to capture the left paralumbar fossa of 20 primiparous cows. For each cow, the system recorded 3D images at 30 frames per second during milking. Each image was automatically processed to locate the left paralumbar fossa region and quantify its average concavity. Then, the average concavity values from all images of 1 cow during 1 milking process were chronologically assembled to form an undulation signal. By applying fast Fourier transformation to the signal, we identified cyclic oscillations that occurred in the same frequency range as reticulo-ruminal contractions. To validate the oscillation identification, 2 trained assessors visually identified reticulo-ruminal contractions from the same 3D image recordings on screen. The matching sensitivity between the automatically identified oscillations and the manually identified reticulo-ruminal contractions was 0.97. This 3D vision system can automate the assessment of reticulo-ruminal motility in dairy cows. It is noninvasive and can be implemented on farms without distressing the cows. It is a promising tool for farmers, giving them regular information about the gastrointestinal health of individual cows and helping them in daily farm management.
[1]
Y. Ruckebusch,et al.
Digestive Physiology and Metabolism in Ruminants
,
1980,
Springer Netherlands.
[2]
U. Braun,et al.
Ultrasonographic and radiographic findings in 503 cattle with traumatic reticuloperitonitis.
,
2018,
Research in veterinary science.
[3]
R. E. Hungate,et al.
The Rumen and Its Microbes
,
2013
.
[4]
T. Bryant,et al.
Diagnosis and Management of Rumen Acidosis and Bloat in Feedlots.
,
2017,
The Veterinary clinics of North America. Food animal practice.
[5]
David Johnson,et al.
Multi‐Modal Sensor Calibration Using a Gradient Orientation Measure
,
2015,
J. Field Robotics.
[6]
Radu Bogdan Rusu,et al.
Semantic 3D Object Maps for Everyday Manipulation in Human Living Environments
,
2010,
KI - Künstliche Intelligenz.
[7]
X Song,et al.
Automated body condition scoring of dairy cows using 3-dimensional feature extraction from multiple body regions.
,
2019,
Journal of dairy science.