Machine vision-based automatic lamb identification and drinking activity in a commercial farm.

Area of Science:

• Agricultural Technology
• Computer Vision
• Animal Science

Background:

• Traditional Radio Frequency Identification (RFID) for lamb ear tag tracking is expensive and impractical for large groups.
• Existing electronic identification systems are not universally mandatory, limiting widespread adoption in some regions.

Purpose of the Study:

• To develop an affordable and easily implementable machine vision system for individual lamb identification using existing ear tags.
• To monitor lamb drinking behavior and collect individual animal data for improved farm management.

Main Methods:

• A machine vision system utilizing an RGB camera and deep learning (You Only Look Once) was developed to detect lamb faces and ear tags.
• Algorithms were implemented for ear tag digit recognition, lamb tracking, and identification number recovery.
• Data was collected from lambs in two pens, with ground truth established through human observation of drinking station visits.

Main Results:

• The system achieved a total accuracy of 93% in identifying individual lambs during drinking events.
• The algorithm successfully identified lambs and recorded their drinking duration in real-time within a natural farm environment.
• The system processed approximately 900 visits to the drinking stations during testing.

Conclusions:

• Machine vision offers a viable, inexpensive alternative to RFID for individual lamb identification in commercial pens.
• This system has significant potential for enhancing farm management through automated data collection and monitoring.
• The proposed system facilitates data-driven decision-making by providing accurate individual animal data.