LastAct: Trajectory-Guided Latest-Activity Localization for Real-Time Smart-Home Activity Recognition
Researchers have developed LastAct, a new framework for real-time human activity recognition in smart homes. This system is designed to handle continuous sensor data streams where activity boundaries are not predefined. LastAct uses trajectory-centric methods and spatial context from floorplans to accurately identify the most recent activity, even within windows containing mixed activities. AI
IMPACT Improves the accuracy and robustness of smart-home activity recognition systems by addressing challenges in real-time data streams.