The case for ambient sensing for human activity detection

Human activity detection using various sources of data is an important problem due to its application in various domains, such as health-care, elderly care, security/safety, etc. Traditionally, this activity detection is carried out using multimedia data, including audio and video resources. Recently, the Internet of Things (IoT) has led to highly-improved computation and communication capabilities even within the smallest devices, giving rise to wearable devices. These devices can collect useful data about movements and thus enable detecting human activities. However, both traditional methods (multimedia) and wearable device-based methods completely expose users, resulting in severe privacy issues. Thus, it is crucial to be able to still detect these activities without compromising the user's privacy. In this paper, we make a case where ambient sensing (sensors that collect data representing only environmental changes, such as temperature, lighting, etc.) can be used to detect human activities. Since the available data corresponds to only the status of the surrounding environment, the user privacy can be preserved. We demonstrate which aspects of ambient sensing methods are desirable and what types of applications can benefit from them.