An Automatic Ear Temperature Monitoring Method for Group-Housed Pigs Adopting Infrared Thermography

Area of Science:

• Animal Science
• Agricultural Engineering
• Biomedical Engineering

Background:

• Monitoring pig health is crucial for early disease detection and welfare.
• Traditional methods for temperature monitoring can be labor-intensive and stressful for animals.
• Infrared thermography (IRT) offers a non-invasive approach for temperature assessment.

Purpose of the Study:

• To develop an automated monitoring system for group-housed pig ear temperature using IRT.
• To implement a neural network for accurate pig ear detection.
• To establish a reliable method for automatic temperature extraction from pig ears.

Main Methods:

• Utilized SwinStar-YOLO neural network for pig ear recognition.
• Employed morphological image processing and temperature matrix for data extraction.
• Processed a dataset of 3600 images (10,812 pig ears) with 5-fold cross-validation.
• Validated temperature readings against manual measurements.

Main Results:

• The ear detection model achieved 93.74% precision (IoU threshold).
• High correlation coefficients (R2) were found between manual and automated temperatures (0.97 max, 0.88 avg).
• The system demonstrated feasibility and reliability for automated pig ear temperature monitoring.

Conclusions:

• The developed automated IRT system is a feasible and reliable tool for monitoring pig ear temperature.
• This technology can serve as a powerful non-invasive method for early health warnings in pigs.
• The AI-driven approach enhances efficiency and accuracy in livestock health management.