Lizard Body Temperature Acquisition and Lizard Recognition Using Artificial Intelligence

Area of Science:

• Animal Biology
• Computer Vision
• Thermal Imaging

Background:

• Manual body temperature acquisition in lab animals lacks accuracy and can cause distress.
• Semi-manual methods using thermal cameras are time-consuming and not fully automatic.

Purpose of the Study:

• To automate the body temperature acquisition process for lizards in laboratory settings.
• To develop a non-invasive and accurate method for monitoring animal physiology.

Main Methods:

• Utilized a dual-camera system (RGB and thermal) for simultaneous data capture.
• Developed an AI model (YOLOv5) for automatic detection and localization of lizard body parts.
• Created a custom dataset of lizard images for training the AI model.
• Implemented coordinate conversion for accurate temperature mapping from thermal to RGB images.

Main Results:

• Achieved high precision (90.00%) and recall (98.80%) in detecting lizards and their body parts using YOLOv5.
• Successfully automated the temperature acquisition process, overcoming challenges of low-contrast environments.
• Enabled accurate temperature readings for specific body parts of the lizard.

Conclusions:

• The developed automated system significantly enhances the accuracy and efficiency of animal body temperature measurement.
• This AI-driven approach offers a less stressful and more reliable alternative to traditional methods for biological research.