Enhancing activity recognition with location related information based on smartwatch sensors

The wearable devices adoption reached a new gap during the last two years. With their form factors, constant proximity with the body and armada of sensors, these devices allow to collect accurate information about our locations, movements and derived activities. Their usage is mainly related to the tracking of basic fitness data: Number of steps, hours of sleep, average heart rate, etc. These devices do not only present opportunities regarding private usage, but companies could also benefit from the collected information from their customers or employees to improve the efficiency and effectiveness of their services. However, the performance of the activity recognition system is nowadays not adapted to complex activities as they are either based on movements or location sensors but not a combination. This paper investigates the opportunities and impact linked to the integration of location related information in human activity recognition system. The performance of the smartwatch as a sensing device has also been analysed. In the first part of this thesis, the background environment related to the smartwatches and activity recognition techniques are presented. A literature review offers then the current state of the techniques and studies about smartwatches, human activity recognition and location related information. This chapter also identifies a research gap about the combination of spatial-time and time series....