Smart Lifelogging: Recognizing Human Activities using PHASOR

Abstract

Abstract:This paper introduces a new idea for sensor data analytics, named PHASOR, that can recognize and stream individual human activities online. The proposed sensor concept can be utilized to solve some emerging problems in smartcity domain such as health care, urban mobility, or security by creating a lifelog of human activities. PHASOR is created from three 'components': ID, model, and Sensor. The first component is to identify which sensor is used to monitor which object (e.g., group of users, individual users, type of smartphone). The second component decides suitable classifiers for human activities recognition. The last one includes two types: (1) physical sensors that utilize embedded sensors in smartphones to recognize human activities, (2) human factors that uses human interaction to personally increase the accuracy of the detection. The advantage of PHASOR is the error signal is inversely proportional to its lifetime, which is well-suited for lifelogging applications. The proposed concept is evaluated and compared to de-facto datasets as well as state-of-the-art of Human Activity Recognition (HAR) using smartphones, confirming that applying PHASOR can improves the accuracy of HAR.

Building a Disclosed Lifelog Dataset

“…Shown in Figure 1 is the data of a logger gathered during 10 years, which signi cantly increases from less than 5 GB with only images and activities logs in 2005 to more than 3.2 TB with rich multi-modal information from images, audio, videos to biometrics in 2015. Captured over a long period of time (e.g., a decade for the logger whose data is reported in Figure 1), heterogeneous lifelogs gathered increasing a ention in recent years within the research community to provide a detailed picture of the experiences of an individual, with numerous applications in terms of assisted technologies for human memory [2], health and wellness [3,4], activities recognition [5], and many others. It is no surprise that lifelogging is also receiving increasing a ention within the research community and is fast becoming a mainstream research topic with the increase of workshops focusing on lifelogs, e.g., NTCIR-12 -Lifelog 2 , ACM MM 2016 -LTA 3 , ImageCLEF 2017 -Lifelog 4 and NTCIR-13 -Lifelog 2 5 .…”

“…Following the spirits of the de nition by Dodge and Kitchen [9], a lifelog typically consists of numerous di erent types of data, such as image/video content from wearable cameras (e.g., SenseCam), audio content from personal audio devices, biometric sensor content from activity trackers (e.g., from a wristband or by a phone as in [5]) or health-monitoring devices, informational content from the media consumed by the lifelogger, and so on. Ideally, we should log all information from all sources, however, it is not doable in practice.…”