Poultry producers could soon trade human oversight for an eye in the sky after researchers successfully used AI-equipped drones to automatically spy on turkeys.
A team led by Penn State achieved a 98 per cent accuracy rate in detecting specific bird actions using commercial drones and machine learning, reports Penn State. The findings appear in the December issue of Poultry Science.
Monitoring the health and welfare of animals on large commercial farms remains a costly and labour-intensive task that typically requires constant human presence.
To automate the process, researchers deployed a small drone equipped with a standard colour camera to record 160 young turkeys at the Penn State Poultry Education and Research Center.
“This work provides proof of concept that drones plus AI can potentially become an effective, low-labour method for monitoring turkey welfare in commercial production,” said Enrico Casella, assistant professor of data science for animal systems at Penn State.
“It lays the groundwork for more advanced, scalable systems in the future.”
Turkey behaviours
The team recorded video four times a day as the birds aged from five to 32 days old.
From these flights, researchers extracted individual image frames for manual labelling of more than 19,000 specific behaviours, including feeding, drinking, sitting, standing, perching, huddling, and wing flapping.
This dataset trained a computer vision model called YOLO (you only look once), an algorithm commonly used for real-time object detection.
The best-performing model correctly identified 87 per cent of all present behaviours and accurately classified specific actions 98 per cent of the time, despite the visually chaotic environment of a real farm.
“The study shows that a drone-equipped AI system can accurately detect turkey behaviours,” said Casella.
“This method could reduce labour demands, it could allow continuous, non-invasive monitoring of bird welfare in commercial farms, and it may also reduce the need for constant human presence, lowering training and staffing burdens.”
