The ability to efficiently track animal behaviour in an animal shelter has direct lifesaving applications. However, monitoring behaviour is a challenge as time and financial resources are often limited, and the size and needs of animal populations within shelters are commonly in flux. Individualized care and early recognition of distress in cats are often missed. Our research required a method of behavioural observation that was simple, accessible, used limited human and computer resources and allowed for real-time feedback. Here, we present BeRSTID, an open-source behaviour real-time spatial tracking identification system demonstrated on six cats in an animal shelter using unique 2D fiducial markers. The markers were attached to custom veterinary paper identification collars for feedback on individual animal behaviour over time. Our findings show BeRSTID correlated closely to human-coded data in both real-time and post-event processing modes of eating and drinking behaviours of cats in naturalistic shelter environments. By building upon a lateral concept of marker tracking for direct applied use in a new context, we present a low-barrier user-friendly solution using common technologies that can track and help improve welfare in animal care facilities such as shelters. Further extensions of BeRSTID can be generalized to track unique subjects in a range of environments for multiple use cases.