Using artificial intelligence to interpret moose
Chen and his collaborator, Virginia Cooperative Extension dairy scientist and associate professor Gonzalo Ferreira, plan to collect audio data from cows, calves and beef cattle in pastures. Machine learning is then used to analyze and catalog thousands of points of acoustic data to interpret cow vocalizations such as mowing, chewing, and burping as signs of stress or disease.
“Think of the baby crying on the plane or in church,” Ferreira says. “As a father, I can tell if a baby is crying because it's hungry or because it wants attention.'' It’s about being able to interpret your needs.”
Chen and Ferreira are particularly interested in identifying the vocal patterns of how cows communicate pain. By analyzing the frequency, amplitude, and duration of cow moans and vocalizations and correlating the audio data with salivary cortisol samples taken from the cows, we can determine whether the cows are unstressed or mildly stressed. , you can categorize whether you are experiencing severe stress and begin to decipher “stress”. language. “
As part of the project, Chen is building a computational pipeline that integrates acoustic data management, pre-trained machine learning models, and interactive visualization of animal sounds. The resulting data will be shared in his open-source web-based application, available to scientists, producers, and the general public. Chen said he hopes this information will guide future protocols to improve animal welfare.
“Anyone can connect and use our model directly to run their own experiments,” he said. “This allows us to convert cow vocalizations into interpretable information that humans can recognize.”