Abstract. In this work, algorithms are developed and evaluated to detect physical activities from data acquired using five small biaxial accelerometers worn simultaneously on different parts of the body. Acceleration data was collected from 20 subjects without researcher supervision or observation. Subjects were asked to perform a sequence of everyday tasks but not told specifically where or how to do them. Mean, energy, frequency-domain entropy, and correlation of acceleration data was calculated and several classifiers using these features were tested. Decision tree classifiers showed the best performance recognizing everyday activities with an overall accuracy rate of 84%. The results show that although some activities are recognized well with subject-independent training data, others appear to require subject-specific training data. The results suggest that multiple accelerometers aid in recognition because conjunctions in acceleration feature values can effectively discriminate many activities. With just two biaxial accelerometers – thigh and wrist – the recognition performance dropped only slightly. This is the first work to investigate performance of recognition algorithms with multiple, wire-free accelerometers on 20 activities using datasets annotated by the subjects themselves. 1

Activity recognition fits within the bigger framework of context awareness. In this paper, we report on our efforts to recognize user activity from accelerometer data. Activity recognition is formulated as a classification problem. Performance of base-level classifiers and meta-level classifiers is compared. Plurality Voting is found to perform consistently well across different settings.

Context-aware computing describes the situation where a wearable / mobile computer is aware of its user’s state and surroundings and modifies its behavior based on this information. We designed, implemented and evaluated a wearable system which can determine typical user context and context transition probabilities online and without external supervision. The system relies on techniques from machine learning, statistical analysis and graph algorithms. It can be used for online classification and prediction. Our results indicate the power of our method to determine a meaningful user context model while only requiring data from a comfortable physiological sensor device. 1.

In recent years much work has been done on human activity recognition using wearable sensors. As we begin to deploy hundreds and even thousands of wearable sensors on regular workers, hospital patients, and army soldiers, the question shifts more toward a balance between what information can be gained and their broad immediate user acceptance. In this paper we compare the activity classification accuracy of four different configurations of accelerometer placement on the human body using hidden Markov models (HMMs). We find the classification accuracy of a single accelerometer placed in three different parts of the body and evaluate whether there is a significant improvement in recognition accuracy by adding multiple accelerometers or not. We also find the number of hidden states that best models each activity by achieving the lowest test error using K-fold cross-validation.

Achieving context-aware computer systems requires that computers can automatically recognize what people are doing. In this work, algorithms are developed and evaluated to detect physical activities from data acquired using five small accelerometers worn simultaneously on different parts of the body. Acceleration data was collected from twenty subjects in both laboratory and semi-naturalistic environments. For semi-naturalistic data, subjects were asked to perform a sequence of everyday tasks outside of the laboratory. Mean, energy, frequency-domain entropy, and correlation of acceleration data was calculated over 6.71 s sliding windows. Decision table, nearest neighbor, decision tree, and Naive Bayesian classifiers were tested on these features. Classification results using individual training and leave-one-subject-out validation were compared. Leave-one-subject-out validation with decision tree classifiers

Abstract. In this paper we demonstrate how a cell phone can infer contextual information such as mode of travel by monitoring the fluctuation of GSM signal strength levels and neighbouring cell information. We show that these signals are stable enough to reliably distinguish between various states of movement such as walking, travelling in a motor car and remaining still. We present preliminary results for a metropolitan environment. 1

Abstract—Post surgical care is an important part of the surgical recovery process. With the introduction of minimally invasive surgery (MIS), the recovery time of patients has been shortened significantly. This has led to a shift of postoperative care from hospital to home environment. To prevent the occurrence of adverse events, the care of these patients is mainly relied on routine visits by home-care nurses. This type of episodic examination can only capture a snapshot of the overall recovery process, and many early signs of potential complication can go undetected. The development of Body Sensor Networks (BSNs) has enabled the use of miniaturised wireless sensors for continuous monitoring of postoperative patients. This paper examines the potential of processing-on-node algorithms for further reducing the wireless bandwidth, and therefore the overall power consumption of the sensors. The accuracy and robustness of the technique are demonstrated with lab experiments and a preliminary clinical case study.

The ability to provide context aware behaviour on a cell phone such as whether a user is walking or driving has previously only been possible via the use of additional hardware sensors such as an accelerometer. In this paper we demonstrate how a level of context awareness similar to that achieved with an accelerometer can be obtained using information readily available on a typical GSM cell phone. We show that by using the patterns of signal strength fluctuations and changes to the current serving cell and monitored neighbouring cells it is possible to distinguish between various states of movement such as walking, driving in a motor car and remaining stationary. We demonstrate how the calibration of the cell phone for use in a given environment can be implemented in an automatic and unsupervised manner, and that we can achieve a classification accuracy of around 80%. 1

The Microsoft SenseCam is a small lightweight wearable camera used to passively capture photos and other sensor readings from a user’s day-to-day activities. It captures on average 3,000 images in a typical day, equating to almost 1 million images per year. It can be used to aid memory by creating a personal multimedia lifelog, or visual recording of the wearer’s life. However the sheer volume of image data captured within a visual lifelog creates a number of challenges, particularly for locating relevant content. Within this work, we explore the applicability of semantic concept detection, a method often used within video retrieval, on the domain of visual lifelogs. Our concept detector models the correspondence between low-level visual features and high-level semantic concepts (such as indoors, outdoors, people, buildings, etc.) using supervised machine learning. By doing so it determines the probability of a concept’s presence. We apply detection of 27 everyday semantic concepts on a lifelog collection composed of 257,518 SenseCam images from 5 users. The results were evaluated on a subset of 95,907 images, to determine the accuracy for detection of each semantic concept. We conducted further analysis on the temporal consistency, co-occurance and relationships

This paper describes a system that can observe, recognise and analyse human movements, to provide this awareness to context-aware applications. The movement recognition and characterisation components of this sentient system are described in detail. The system uses the ground reaction force to classify and analyse movements in an non-clinical environment. The signal is classified using statistical pattern recognition. Equipped with knowledge of the movement, characterisation is the process of analysing the ground reaction force to extract parameters of the movement. The components of the movement awareness system operate in a distributed computing environment.