AI-Powered Surveillance Tracks Mithun Behaviour at Nagaland Research Farm
ICAR scientists in Nagaland use AI and CCTV to automatically detect and track Mithun behaviour, aiding health and breeding management.
Researchers at the ICAR-National Research Centre on Mithun in Nagaland have developed an artificial intelligence system that can automatically detect and track the behaviour of Mithun (Bos frontalis) in real time. The non-contact framework uses cameras and computer vision to monitor animals continuously, offering a potential alternative to labour-intensive manual observation.
The study, published in the journal Engineering Research Express, focused on four key behaviours: feeding, standing, lying and mounting. Twelve high-definition CCTV cameras, including infrared night-vision capability, were installed across two sheds at the institute's farm. The team compiled a dataset of 3,000 manually annotated images to train the AI model.
The system combines the YOLOv8n object-detection model with DeepSORT tracking technology to identify behaviours and assign persistent identities to individual animals across video frames. It achieved a mean average precision of 99.5 per cent and a recall of 99.6 per cent, processing about 31 frames per second on a standard GPU. The framework performed reliably under challenging conditions such as partial occlusion, motion blur, shadows and low-light infrared footage.
Mithun, often called the 'Cattle of the Hills', holds deep cultural and economic significance for tribal communities in Northeast India. Changes in feeding, standing or lying patterns can signal health or comfort issues, while mounting behaviour is relevant for breeding and oestrus detection. Automated monitoring could allow farmers to access such insights without round-the-clock physical presence.
The researchers note that the system has so far been validated only at a single farm. Further testing across different locations, seasons and herd densities is needed. They also plan to expand the framework to recognise additional behaviours like aggression, grooming and disease-related inactivity, and to explore edge-device deployment for wider practical use.