Implementation and Evaluation of a Low-Power Sound-Based User Activity Recognition System

Stäger, Mathias; Lukowicz, Paul; Tröster, Gerhard

doi:10.1109/iswc.2004.25

Cited by 42 publications

(25 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For the ASIC, it would be above 100 classifications @1 MHz, whilst on an XScale processor performing around more than 100 classifications @150 MHz (minimum clock frequency) can be achieved. This can be attributed due to the lower complexity of the decision tree classifier in comparison to a K-NN classifier algorithm [4,6]. The active energy costs (E Nb ) of all the systems are calculated to perform 100 classifications for task B.…”

Section: Case Study: Discussionmentioning

confidence: 99%

“…This vision implies that the wearable sensor nodes should be extremely small and consume so little power that no power source change is required for several months to years. Working towards this vision, the previous work of our group dealt with issues such as, activity recognition using low-power features and classifier algorithms [4,6], optimization of power and size in a multi-sensor context recognition platform [7], development of hybrid micro power supply to achieve autonomous behavior [8], electronic packaging aspects of an ultra-miniaturized wearable sensor button, reliability modeling of embedded systems in wearable computing [9,10], detailed systematic approach considering wearability and power consumption [11] and methodologies for context-aware system design were proposed [12] for selecting optimized architectures with respect to power consumption. The main aspect which sets us aside from the work done by other groups in the field of personal and ubiquitous systems is the focus on context aware wearable systems.…”

Section: Related Work and Paper Contributionmentioning

confidence: 99%

“…It is assumed that we have a priori knowledge about what sensors are required in each activity. The features and the classifier algorithms are known [4,6]. They range from simple daily-life activities detection using 'mean' feature with a C 4.5 decision tree classifier algorithm to solving a complex health monitoring task using Hidden Markov Models (HMM).…”

Section: Tasksmentioning

confidence: 99%

See 2 more Smart Citations

Functionality-power-packaging considerations in context aware wearable systems

Bharatula

Lukowicz

Tröster

2006

Pers Ubiquit Comput

View full text Add to dashboard Cite

Wearable computing places tighter constraints on architecture design than traditional mobile computing. The architecture is described in terms of miniaturization, power-awareness, global low-power design and suitability for an application. In this article we present a new methodology based on three different system properties. Functionality, power and electronic Packaging metrics are proposed and evaluated to study different trade offs. We analyze the trade offs in different context recognition scenarios. The proof of concept case study is analyzed by studying (a) interaction with household appliances by a wrist worn device (acceleration, light sensors) (b) studying walking behavior with acceleration sensors, (c) computational task and (d) gesture recognition in a wood-workshop using the combination of accelerometer and microphone sensors. After analyzing the case study, we highlight the size aspect by electronic packaging for a given functionality and present the miniaturization trends for 'autonomous sensor button'.Keywords Wearable computing Á Context recognition Á Gesture Á Electronic packaging Á Functionality Á Miniaturization Context aware wearable systemsWearable computing as defined by [1, 2] envisions personal, mobile computing systems that are always on, useful in all situations and most of all, easy to use. Thus whereas a conventional mobile device would only be used for an occasional schedule check or address lookup, a wearable device would constantly provide the user with useful information such as nearby shops and special offers, transport delays, or health and lifestyle-related reminders (taking medicine, diet etc). Such systems are particularly important in professional applications such as emergency response units, manufacturing and maintenance. Thus a wearable system might constantly provide a fireman with hints and warning about hazards related to his environment, his physiological state and his current actions.A key component of the wearable computing vision is the ability of the system to model and recognize user activity and the situation around him. This so called context awareness [3] allows the system to proactively provide the user with the right information at the right time, reduces the complexity of the user interface, and allows new modes of information recording. One of the most popular approaches to context awareness in a mobile environment is based on simple on-body sensors. Thus an accelerometer, light sensor and a microphone placed on the wrist could be used to track interaction with household appliances [4] or the use of tools [5]. In a similar way an accelerometer and/or gyroscope on the upper leg can differentiate between level walking, going upstairs, going downstairs and running.

show abstract

Section: Case Study: Discussionmentioning

confidence: 99%

Section: Related Work and Paper Contributionmentioning

confidence: 99%

Section: Tasksmentioning

confidence: 99%

See 1 more Smart Citation

Functionality-power-packaging considerations in context aware wearable systems

Bharatula

Lukowicz

Tröster

2006

Pers Ubiquit Comput

View full text Add to dashboard Cite

show abstract

“…The paper discusses a prototype of a sound recognition system focused on an ultra low power hardware implementation in a button-like miniature form. The implementation and evaluation of the final version of the prototype are performed in [SLT04]. In those papers, the authors have used FFT features and compared a k-nearest centre classifier with a k-nearest neighbour classifier.…”

Section: Audio Recognition For a Given Environmentmentioning

confidence: 99%

Acoustic Event Detection and Classification

Temko

Nadeu

Macho

et al. 2009

Computers in the Human Interaction Loop

View full text Add to dashboard Cite

The human activity that takes place in meeting-rooms or class-rooms is reflected in a rich variety of acoustic events, either produced by the human body or by objects handled by humans, so the determination of both the identity of sounds and their position in time may help to detect and describe that human activity. Additionally, detection of sounds other than speech may be useful to enhance the robustness of speech technologies like automatic speech recognition.Automatic detection and classification of acoustic events is the objective of this thesis work. It aims at processing the acoustic signals collected by distant microphones in meeting-room or classroom environments to convert them into symbolic descriptions corresponding to a listener's perception of the different sound events that are present in the signals and their sources.First of all, the task of acoustic event classification is faced using Support Vector Machine (SVM) classifiers, which are motivated by the scarcity of training data. A confusion-matrix-based variable-feature-set clustering scheme is developed for the multiclass recognition problem, and tested on the gathered database. With it, a higher classification rate than the GMM-based technique is obtained, arriving to a large relative average error reduction with respect to the best result from the conventional binary tree scheme. Moreover, several ways to extend SVMs to sequence processing are compared, in an attempt to avoid the drawback of SVMs when dealing with audio data, i.e. their restriction to work with fixed-length vectors, observing that the dynamic time warping kernels work well for sounds that show a temporal structure. Furthermore, concepts and tools from the fuzzy theory are used to investigate, first, the importance of and degree of interaction among features, and second, ways to fuse the outputs of several classification systems. The developed AEC systems are Two kinds of databases are used: two databases of isolated acoustic events, and a database of interactive seminars containing a significant number of acoustic events of interest. Our developed systems, which consist of SVM-based classification within a sliding window plus post-processing, were the only submissions not using HMMs, and each of them obtained competitive results in the corresponding evaluation.vi Speech activity detection was also pursued in this thesis since, in fact, it is a -especially important -particular case of acoustic event detection. An enhanced SVM training approach for the speech activity detection task is developed, mainly to cope with the problem of dataset reduction.The resulting SVM-based system is tested with several NIST Rich Transcription (RT) evaluation datasets, and it shows better scores than our GMM-based system, which ranked among the best systems in the RT06 evaluation.Finally, it is worth mentioning a few side outcomes from this thesis work. As it has been carried out in the framework of the CHIL EU project, the author has been responsible for the organization of the above m...

show abstract

“…used simple signal processing to measure activity levels of users wearing a mobile device. Several methods for low power embedded context classification have been introduced in the community [3] [4][5] [6], and trade-offs that must be made between classification quality and energy consumption for embedded context recognition are discussed in [7], [8].…”

Section: Introduction and Related Workmentioning

confidence: 99%

Using prediction to conserve energy in recognition on mobile devices

Gordon

Sigg

Ding

et al. 2011

2011 IEEE International Conference on Pervasive Computing and Communications Workshops (PERCOM Workshops)

View full text Add to dashboard Cite

Abstract-As devices are expected to be aware of their environment, the challenge becomes how to accommodate these abilities with the power constraints which plague modern mobile devices. We present a framework for an embedded approach to context recognition which reduces power consumption. This is accomplished by identifying class-sensor dependencies, and using prediction methods to identify likely future classes, thereby identifying sensors which can be temporarily turned off. Different methods for prediction, as well as integration with several classifiers is analyzed and the methods are evaluated in terms of computational load and loss in quality of context. The results indicate that the amount of energy which can be saved is dependent on two variables (the acceptable loss in quality of recognition, and the number of most likely classes which should be accounted for), and two scenario-dependent properties (predictability of the context sequences and size of the context-sensor dependency sets).

show abstract

Implementation and Evaluation of a Low-Power Sound-Based User Activity Recognition System

Cited by 42 publications

References 8 publications

Functionality-power-packaging considerations in context aware wearable systems

Functionality-power-packaging considerations in context aware wearable systems

Acoustic Event Detection and Classification

Using prediction to conserve energy in recognition on mobile devices

Contact Info

Product

Resources

About