The Ministry of Agriculture & Farmers Welfare has announced an artificial intelligence system developed to monitor the behaviour of Mithun, the cattle breed traditionally raised in the northeastern hills. Researchers at the ICAR-National Research Centre on Mithun in Nagaland have created this technology to track animals in real time without requiring constant human observation. The system could eventually help farmers manage animal health, breeding and welfare more effectively.
How the system works
The research team set up 12 high-definition CCTV cameras across farm sheds, with both daytime and infrared night-vision capabilities. Using 3,000 manually annotated images, they trained an artificial intelligence model to identify four key animal behaviours: feeding, standing, lying down and mounting. The system combines detection technology with tracking software that assigns unique identities to individual animals and follows them across video footage.
The AI framework achieved a 99.5 percent accuracy rate in identifying behaviour and processed video at 31 frames per second. It successfully tracked animals even in difficult conditions such as partial obstruction, wet ground, shadows and nighttime footage using infrared cameras. The system detected mounting behaviour at 97 percent confidence in infrared conditions, demonstrating its capability for round-the-clock monitoring.
Mithun, known locally as the “Cattle of the Hills”, holds cultural and economic importance for tribal communities across Northeast India. The breed contributes significantly to food security and livelihoods in the region.
What this means for you
Farmers raising Mithun could eventually access real-time information about animal health and behaviour without spending hours observing their herds manually. Changes in eating, movement and posture patterns can signal health problems or reproductive readiness, allowing farmers to act quickly. This technology could particularly benefit farmers who manage large herds or need to monitor animals during night hours when traditional observation is impractical.
The researchers acknowledge that the system requires further testing across different farms, regions and seasons before wider adoption. They plan to expand the technology to recognise additional behaviours such as disease signs, aggression and grooming patterns in future work.