Towards Automated Prototyping of Gesture Recognition Systems for Wearable Devices using Inertial Sensors

Inhalt:
Embedded always-on gesture recognition systems demand for a specific trade-off in terms of computational load, memory footprint, and accuracy. Hence, special care has to be taken in choosing the right recognition algorithms. But even with a suitable recognition algorithm, gesture development is still a manual, labor-intensive, and error-prone task, deteriorating the overall design quality. In this paper, we propose a framework for automated prototyping of embedded gesture recognition systems for mobile and wearable devices using inertial sensors. First results show the viability and scalability of the approach, which might serve as a basis of future automatic design of sensor-based embedded gesture recognition systems.