Kohima: Researchers at the ICAR-National Research Centre on Mithun in Nagaland built an AI system to monitor Mithun behavior. The project tracks these animals in real time to help with breeding and health. They installed 12 high-definition CCTV cameras across two farm sheds. These provide around-the-clock surveillance and include infrared coverage.
The team created a dataset containing 3,000 manually annotated images. These images show four specific behaviors: feeding, standing, lying, and mounting. The system uses the YOLOv8n model to spot these actions. It also employs DeepSORT technology to give every animal a persistent ID. During testing, the YOLOv8n model hit 99.5 percent mean average precision and 99.6 percent recall on an NVIDIA RTX 3060 GPU.
The tech faces tough conditions. It handles shadows, wet ground, and motion blur. Investigators recorded lying and feeding behaviors at 96 percent confidence. They spotted mounting behavior at 97 percent confidence using infrared night vision. Researchers noted: "Mithun (Bos frontalis), popularly known as the 'Cattle of the Hills', has significant social, cultural and economic importance for tribal communities across Northeast India."
The system is currently limited to one farm. It needs more testing across different seasons and regions to prove reliability. Severe crowding or blocked views can still throw off the sensors. Future plans include identifying aggression, grooming, and signs of sickness. The study involved teams from ICAR-NRC on Mithun, NIT Nagaland, Nagaland University, and CHRIST (Deemed to be University). It appears in Volume 8 of Engineering Research Express.
Comments