**2. Related Work**

In the research field of Human Activity Recognition (HAR), there has been a growing trend of wrist-worn wearable devices (e.g., smart watches) with Inertial Measurement Units (IMUs) containing a host of different sensors. This was highlighted by a recent survey conducted in [9], which showed an even bigger trend in the use of IMUs with accelerometers, gyroscopes and magnetometers. The increased usage and interest of wrist-worn devices is not surprising, given the global acceptance of these devices in our daily lives. According to Statista [6], the number of worldwide sales for smart watches was five million in 2014, predicted to exceed 75 million in 2017, an increase of 1500% in three years.

There has also been a large amount of research conducted in the field of HAR that uses body-worn sensors, as highlighted by [10]. Body-worn approaches have been shown to have slightly better accuracy [11–14] than approaches that use only wrist-worn devices. However, body-worn approaches currently have the draw-back of being more obtrusive than their wrist-worn counterparts. Prolonged use of body-worn wearable devices may interfere with users' daily lives, thus interfering with how the user performs activities. Moreover, body-worn wearables may become uncomfortable, limiting

the devices to only being worn for short periods of time, rendering them less viable for monitoring daily activities. Wrist-worn approaches are less obtrusive and have less impact on the users' daily lives, allowing the devices to be worn for longer periods of time. The literature presented is primarily focused on recent work that involves the use of wrist-worn wearables in the context of HAR.

We have provided some key details of related work in this area, as shown in Table 1. The details given in the table are:


Furthermore, we classify captured activities into two general categories, low and high level activities. Low level activities are the ones that require whole body movement in order to be performed, such as walking, running and jumping. High level activities involve interaction with objects as described by [9] and daily activities being performed, as categorised by [10].
