Sensor-based human activity recognition for low-power wearable devices
Deadline: 06/06/2026
Scientific-Disciplinary Group
09/IINF-05 - Information Processing Systems
Description
In particular, the research project proposes the design and ondevice validation of lightweight neural network architectures capableof recognizing daily activity from wearable accelerometer and gyroscope data. The methodology integrates deep-learning technologies with model compression techniques, including quantization, pruning, and knowledge distillation, to enable on-device inference within a low power budget. The proposed models will be validated on a publicly available benchmark dataset and compared with the state-of-the-art performances. In parallel, the energy consumption of different wearable devices will be captured using a measurement setup to evaluate a tradeoff between classification accuracy and energy sustainability.
Job posting website
Number of positions
1
Funding body
Università degli Studi di Urbino Carlo Bo
View the original posting on the MUR website: Go to MUR website