Title: Movement Visualization and Activity Recognition using Smart Wearables
Abstract:
Automatic recognition of human activities is attributed with great importance for its widespread and potential applications in different domains, including healthcare, safety, behavior monitoring, energy management, manufacturing, and elderly care. Human Activity Recognition (HAR) is the cornerstone for most of these applications, and so the performance of such an application largely depends on the accuracy and robustness of the underlying activity recognition models. However, activity recognition is challenging, particularly in natural settings, due to issues like confounding gestures present in different activities, diversity in performing the same activity, and a wide range of possible human activities. Also, on-device processing, required by many real-time applications, is challenging due to limited resources available in the wearable devices. In contrast to state-of-the-art methods that mostly emphasize on feature engineering and classification techniques for HAR, our works focus on the representation and manipulation of device orientations, and we developed a set of solutions with improved accuracy and reduced computation overhead. Understanding data and their characteristics is fundamental toward developing data-driven solutions. We developed a novel method for visualizing movement and orientation using inertial sensors. Our visualization method is complementary to the existing methods, and it provides additional utility toward better understanding the movements associated with different activities. We developed a system to monitor family eating dynamics that can monitor and measure eating behaviors and their complex determinants in real-time and in context. As part of the system, we built an efficient and effective solution for eating detection leveraging the orientation of the wrist. Motivated by the importance of hand hygiene compliance in healthcare settings and food businesses as well as in daily life, we developed a robust solution for handwash detection that can be used to provide reminders when a user forgets to wash hands. This thesis presents an interactive reminder system that overcomes some limitations of smartwatches. Finally, we present an easy-to-use app for sensor data collection using smartwatches. The app facilitates prompt data collection without requiring expertise and effort for custom device/app development.
Committee:
- Jack Stankovic (Advisor)
- Alf Weaver (Committee Chair)
- Jack Davidson
- Laura Barnes
- John Lach (The George Washington University)