Early morning activity detection using acoustics and wearable wireless sensors

In this paper, we study the classification of the early morning activities of daily living to assist those with cognitive impairments due to traumatic brain injuries. The system can be used to help therapists in hospitals or could be deployed in one's home. We briefly describe the infrastructure of our cost-effective system which uses fixed and wearable wireless sensors and show results related to the detection of activities executed in the morning. We focus on several early morning activities such as washing face, brushing teeth, shaving face and especially on the detection of shaving and brushing activities executed with electric devices. Features from accelerometer and acoustic sensors were extracted in time and frequency domain then used for classification using Gaussian mixture models, followed by a sequential classifier. We show promising classification results obtained from 7 subjects especially when electric devices are used to execute these early morning activities.