Smoking recognition with smartwatch sensors in different postures and impact of user's height

Currently, smartwatches are mainly used as an extension of smartphones. However, equipped with various motion sensors, they are also effective devices for human activity recognition, particularly for those involving hand and arm movements. In this paper, we investigate the smoking recognition problem with motion sensors on smartwatches using supervised learning algorithms. For this purpose, we collected a dataset from 11 participants including ten different activities. The dataset includes different smoking variations in four different postures, such as smoking while standing, as well as similar activities, such as eating, and other activities, such as walking. Instead of approaching the problem as a binary classification problem, such as smoking and other, we are interested in differentiating smoking in different postures. Our aim is to explore the parameter space that may affect the recognition process on a large and complex dataset, considering 4 different window sizes and overlaps, 63 different features extracted from each sensor, 4 different sensors, 2 different sensor combinations, 3 classifiers and 10 different activities. Additionally, we analyze the impact of participants' height on the recognition performance. The results show that, simple time-domain features and the combination of accelerometer and gyroscope sensors perform the best. When we consider the impact of height on the recognition performance, the results show that it does not have a significant effect when all activities are considered, however, it does have an effect on smoking while standing, particularly for participants with a significant height difference than others.