Deep Learning Classification of Canine Behavior Using a Single Collar-Mounted Accelerometer: Real-World Validation

Abstract

Collar-mounted canine activity monitors can use accelerometer data to estimate dog activity levels, step counts, and distance traveled. With recent advances in machine learning and embedded computing, much more nuanced and accurate behavior classification has become possible, giving these affordable consumer devices the potential to improve the efficiency and effectiveness of pet healthcare. Here, we describe a novel deep learning algorithm that classifies dog behavior at sub-second resolution using commercial pet activity monitors. We built machine learning training databases from more than 5000 videos of more than 2500 dogs and ran the algorithms in production on more than 11 million days of device data. We then surveyed project participants representing 10,550 dogs, which provided 163,110 event responses to validate real-world detection of eating and drinking behavior. The resultant algorithm displayed a sensitivity and specificity for detecting drinking behavior (0.949 and 0.999, respectively) and eating behavior (0.988, 0.983). We also demonstrated detection of licking (0.772, 0.990), petting (0.305, 0.991), rubbing (0.729, 0.996), scratching (0.870, 0.997), and sniffing (0.610, 0.968). We show that the devices’ position on the collar had no measurable impact on performance. In production, users reported a true positive rate of 95.3% for eating (among 1514 users), and of 94.9% for drinking (among 1491 users). The study demonstrates the accurate detection of important health-related canine behaviors using a collar-mounted accelerometer. We trained and validated our algorithms on a large and realistic training dataset, and we assessed and confirmed accuracy in production via user validation.

Effect of Music on Stress Parameters in Dogs during a Mock Veterinary Visit

“…Small devices such as accelerometers are also used in the pet industry. They commonly attach to collars and accurately capture location and movement and are being developed further using machine learning to accurately detect other behaviors [ 69 ]. Recent advances in pet technology mean that these types of devices, along with other types of wearables (harnesses, etc.…”

Determination of the center of mass in a heterogeneous population of dogs

“…We elected to use an IMU box on the ventral part of the neck of the dog as a point of reference because it is a common place for activity monitors to be placed in clinical trials [ 20 – 23 ]. Inertial measurement units are motion-based sensors that offer an opportunity to monitor the activity of a canine patient in their natural environment; information that would be useful for determining the impact of disease burden and treatments for osteoarthritis, chronic pain, cardiovascular disease, and obesity.…”

Deep Learning Empowered Wearable-Based Behavior Recognition for Search and Rescue Dogs

“…These results helped the current work to decide the optimal sensor placement, which was mounting a harness on the back of the SaR dog with the developed device in it. Finally, the authors in [29] created a huge dataset exploiting a 3-axial accelerometer and collecting data from more than 2500 dogs of multiple breeds. Then they trained a deep learning classifier which was then validated for a real-world detection of eating and drinking behavior.…”