AI Mithun Monitoring: Can Cameras Help Farmers Track Animal Health and Breeding?

Mithun is an important livestock species for communities across Northeast India. But farmers managing free-ranging animals can find it difficult to observe every animal throughout the day. ICAR researchers in Nagaland are now testing artificial intelligence to monitor Mithun behaviour continuously using cameras installed on the farm.

The system was developed at the ICAR-National Research Centre on Mithun in Nagaland. Researchers used 12 high-definition CCTV cameras across two sheds, including infrared coverage for nighttime monitoring. The system can identify four behaviours: feeding, standing, lying and mounting.

The researchers created a dataset containing 3,000 manually annotated images of Mithun. These images were used to train an AI system to recognise animal behaviour. Another system tracks individual animals between video frames, allowing researchers to follow their movements over time.

This approach could help farmers notice changes that may otherwise be missed during routine observation. A reduction in feeding or unusual changes in activity can sometimes indicate that an animal needs attention, although the technology cannot by itself diagnose a disease.

How Can It Help Farmers?

The AI system has two main functions. The first identifies what a Mithun is doing, while the second follows individual animals. Together, these systems can provide continuous information about behaviour without requiring a farmer or livestock worker to watch the herd throughout the day.

The researchers reported a mean average precision of 99.5% at the mAP@0.5 measure for the behaviour-detection model, with recall reaching 99.6%. Detection operated at about 31 frames per second using the tested computer hardware, allowing the system to work in real time.

For livestock management, behaviour can provide useful information about animal condition. Feeding, standing and lying patterns can change when animals experience discomfort, illness or other problems. Farmers could potentially use such changes as an early signal to inspect an animal more closely.

Mounting behaviour can also have reproductive importance. Changes in mounting activity may provide information relevant to oestrus and breeding management. Detecting such behaviour automatically could become useful where farmers have difficulty observing animals continuously.

The system also works with nighttime infrared footage. This is important because livestock do not stop moving or displaying behaviour after sunset. Continuous monitoring could therefore provide information that ordinary daytime observation may miss.

Why Mithun Farmers Could Benefit

Mithun is particularly associated with the hill and forested regions of Northeast India, where animals can be managed under conditions very different from intensive dairy farms. Farmers may have to cover large areas to check animals, making frequent physical monitoring difficult.

ICAR says the new technology could eventually help with breeding, animal health and round-the-clock monitoring. A farmer could potentially receive information about an animal showing unusual activity without physically observing the herd at every moment.

The technology may also help researchers understand normal Mithun behaviour. Once larger datasets are created, AI systems could potentially be trained to recognise additional behaviours such as aggression, grooming or periods of reduced activity.

This could be particularly useful for identifying animals that require closer observation. But farmers should understand that an AI alert would be a signal for inspection, not a veterinary diagnosis. Any suspected illness or reproductive problem should still be assessed by a qualified veterinarian or livestock expert.

Technology Is Still Being Tested

The current system is not yet a ready-made product that every Mithun farmer can install. ICAR researchers have tested it on a single farm, and its performance still needs to be evaluated across different farms, seasons, animal densities and camera arrangements.

The researchers also noted challenges caused by animals blocking one another, background clutter, uneven ground, shadows and other conditions. Such problems can affect the ability of a camera-based system to identify and track individual animals accurately.

Another limitation is that the present system recognises only four behaviours. More work is required before it can reliably identify a wider range of health or behavioural indicators. Researchers are looking at larger datasets and additional AI models for future development.

The cost of cameras, computing equipment, electricity, internet connectivity and maintenance would also matter if the system moves from research farms to commercial livestock holdings. Farmers would need a system designed for local conditions rather than simply transferring laboratory equipment to the field.

Could It Work Beyond Mithun?

The basic idea has wider potential. Dairy cattle, buffaloes, sheep and goats also display behavioural changes that can provide information about feeding, movement, reproduction and health. Camera-based monitoring could eventually become one part of precision livestock management.

But each species behaves differently, and AI models trained on one animal cannot automatically be assumed to work accurately on another. New datasets and field testing would be required for cattle, buffaloes, goats or sheep.

For farmers, the biggest potential benefit is continuous observation. A camera does not replace the farmer or veterinarian, but it can collect information throughout the day and provide an additional way to identify animals that deserve attention.

The technology could also become more practical as smaller computing devices and edge-based AI systems improve. If data can be processed locally without expensive equipment or continuous internet connectivity, livestock monitoring may become easier to use in remote farming areas.

Mithun farmers should not rush to purchase untested AI monitoring equipment. The current ICAR system remains a research development, and its performance needs wider validation before farmers can judge its cost and practical benefits.

Farmers interested in livestock monitoring can instead begin with basic observation records. Recording feeding, movement, reproductive behaviour, illness and treatment can help identify changes in individual animals and provide useful information to veterinarians.

For farmers with larger herds, simple cameras may already help improve observation around sheds and feeding areas. But any technology should be selected according to herd size, farm layout, electricity availability, connectivity and the farmer’s ability to respond to alerts.

Can AI Change Mithun Farming?

ICAR’s Mithun project shows how artificial intelligence could become an additional tool for livestock management. The technology can watch animals continuously and identify behaviours that may provide clues about health, comfort and reproduction.

It is still early research, not a replacement for farmers or veterinarians. The next step is wider testing across farms and seasons and the development of systems that are affordable and practical for livestock keepers.

For Mithun farmers in remote regions, the idea is simple: instead of relying only on someone watching the herd, technology could provide another pair of eyes that works day and night.