Canine activity monitoring system using wearable collar device

Abstract

This paper presents a real-time pet activity monitoring system. With the societal shift recognizing pets as important members enriching family life, there is a growing interest in understanding the health and emotional state of pets through behavioral analysis. However, continuously observing pets is often challenging for pet owners due to their busy lives. To address this issue, we propose a wearable collar device that uses accelerometers and gyroscopes to collect motion data of pets. The collected data undergoes a calibration process for gravity and is then processed and analyzed using a ResNet model to classify pet behaviors. The system we propose enables pet owners to monitor pet behavior anytime and anywhere, allowing for early detection of abnormal behavior or health issues and appropriate responses. This research provides detailed explanations of the overall system operation, device design, data collection, and labeling, and the method of behavior classification, along with a real-time application.