Refining animal care through technology: Addressing alopecia in Jaculus jaculus with validated computer vision...

Area of Science:

• Ethology
• Animal Welfare Science
• Computer Vision Applications

Background:

• Lesser Egyptian Jerboas (Jaculus jaculus) exhibit rapid, difficult-to-observe movements, complicating behavioral assessment.
• Alopecia and excessive jumping in captive jerboas necessitated a more robust welfare evaluation method.
• Traditional behavioral observation methods struggle with short-duration events.

Purpose of the Study:

• To validate an open-source computer vision toolkit for analyzing jerboa behavior and assessing welfare.
• To compare the accuracy and precision of human observers versus computer vision algorithms in behavioral classification.
• To investigate the relationship between behavior, enrichment, and alopecia in captive jerboas.

Main Methods:

• Comparison of human observers and machine learning-based computer vision algorithms for behavioral classification.
• Validation of computer vision accuracy and precision against human-graded standards.
• Annotation of captive jerboa activity budgets using computer vision before and after enrichment implementation.

Main Results:

• Computer vision achieved human-grade accuracy and intraclass correlation coefficient (ICC) in behavioral classification, outperforming humans with short-duration behaviors.
• Alopecia was significantly correlated with grooming behavior.
• Enrichment strategies, including terrarium height and opaque dividers, influenced grooming, but conventional enrichment did not significantly alter overall behavior.

Conclusions:

• Computer vision provides a fast, accurate method for analyzing complex animal behavior, crucial for welfare studies.
• The study suggests psychogenic alopecia may occur in captive jerboas, linked to grooming behaviors.
• Automated behavioral analysis using computer vision can inform species-specific husbandry and improve animal welfare.