Dynamic multi-component based activity detection and recognition within smart homes

Within smart homes, ambient sensors are used to monitor the interaction between users and the home environment. The data produced from the sensors is used as the basis for the inference of the users' behaviour information. Partitioning sensor data in response to individual instances of activity is critical for a smart home to be fully functional and to fulfil its role, such as correctly measuring health status and detecting emergency situations. In this study, we propose a multi-component based sensor segmentation approach in an effort to detect and recognise activities within a smart home. A publicly available dataset containing 1,230 sensor events generated in response to 245 instances of 7 activity classes along with 59 idle periods was used to validate the approach. Initial results of the proposed methods have been encouraging achieving 95.72%, 89.52% and 86.13% in respect to precision, recall and accuracy in detecting and recognising activities in addition to the capability of detecting interleaved activities.