Behaviour 2019
Pilot study of the use of barcode detection for cat behaviour monitoring in an animal shelter
Bailey H Eagan1, Emilia Gordon2, Ben Eagan3, Protopopova Alexandra1. 1University of British Columbia, Vancouver, BC, Canada; 2British Columbia Society for the Prevention of Cruelty to Animals, Vancouver, BC, Canada; 3Independent Researcher, Toronto, ON, Canada

Monitoring behaviour in animal shelters is critical for ensuring health and welfare. Computer vision technology may be used as an accessible and low-cost behavioural data collection method to monitor behaviour automatically. We used the OpenCV fiducial tag library ArUco to generate unique barcodes printed on breakaway collars, which created identifiers recognized in video output. Two cameras placed within the cat housing room, including above the litter box and food and water bowls, recorded cats (n=3) continuously for 99 hours. The video streams were subjected to code that captured individual IDs and timestamps within defined regions. A trained observer also coded the videos. Cohen's kappa was run to determine if there was an agreement between human observation and video for the presence or absence of a cat. This automated method's validity is evidenced by preliminary data demonstrating high accuracy between the two methods (κ = 0.928, n=1). These preliminary results suggest that computer vision techniques may be beneficial for shelter research and animal care. This will be confirmed by ongoing data collection and analysis including additional cats that will be presented.