A Machine Learning Approach to Detect the Dog's Behavior using Wearable Sensors

In recent years welfare of animals is the biggest challenge because animals, especially dogs are widely recognized as pet as well as they are using as service animals. So, for the wellbeing of the dog it is necessary to perform objective assessment to track their behavior in everyday life. In this paper, we have proposed an automatic behavior assessment system for dogs based on a neck worn and tail worn accelerometer and gyroscope platform, and data analysis techniques that recognize typical dog activities. We evaluate the system based on the analysis of 8 behavior traits in 3 dogs, incorporating 2 breeds of various sizes. Our proposed framework able to reproduce the manual assessment that is based on the video recording which is treated as gold standard that exhibits the real-life use case of automated dog behavior analysis.