The DOI matches its Crossref record; 22 of the 22 references it cites were found in a registry or an index.
Deep learning for human activity recognition: A resource efficient implementation on low-power devices
Abstract
Human Activity Recognition provides valuable contextual information for wellbeing, healthcare, and sport applications. Over the past decades, many machine learning approaches have been proposed to identify activities from inertial sensor data for specific applications. Most methods, however, are designed for offline processing rather than processing on the sensor node. In this paper, a human activity recognition technique based on a deep learning methodology is designed to enable accurate and real-time classification for low-power wearable devices. To obtain invariance against changes in sensor orientation, sensor placement, and in sensor acquisition rates, we design a feature generation process that is applied to the spectral domain of the inertial data. Specifically, the proposed method uses sums of temporal convolutions of the transformed input. Accuracy of the proposed approach is evaluated against the current state-of-the-art methods using both laboratory and real world activity datasets. A systematic analysis of the feature generation parameters and a comparison of activity recognition computation times on mobile devices and sensor nodes are also presented.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
In the citing paper
“The proposed method results in an overall accuracy of 92.6%, outperforming other methods such as HMMs [23], DBNs [6], and CNNs [11], as shown in Figure 12b.”
From Deep Recurrent Neural Networks for Human Activity Recognition (2017) · cites this paper as [11]
In the citing paper
“Ravi et al. [170] used a shallow three-layer CNN network including a convolutional layer, a fully connected layer, and a softmax layer to perform on-device activity recognition on a resource-limited platform and shown its effectiveness and efficiency on public datasets.”
From Deep Learning in Human Activity Recognition with Wearable Sensors: A Review on Advances (2022) · cites this paper as [170]
In the citing paper
“The main disadvantage of AIDS is the false alarm rate (FAR), making it difficult to distinguish between normal and abnormal intrusion detection patterns [197].”
From Comparative Analysis of Intrusion Detection Systems and Machine Learning Based Model Analysis through Decision Tree (2023) · cites this paper as [197]
Citations
Cites
The Design and Implementation of FFTW3
Frigo, M., & Johnson, S. G. (2005) · Proceedings of the IEEE · doi:10.1109/jproc.2004.840301 · Cited by 5,086 · cited 1×
“For both architectures, we use the FFTW3 library [17] to extract the spectrogram and the Torch framework [18] to implement the deep learning model.”
MethodActivity recognition using cell phone accelerometers
Kwapisz, J. R., Weiss, G. M., & Moore, S. A. (2011) · ACM SIGKDD Explorations Newsletter · doi:10.1145/1964897.1964918 · Cited by 2,912 · cited 1×
“1 6 29 Front trouser pocket 20Hz 1,098,207 [19] (thigh) Daphnet FoG 2 10 Trunk, thigh, ankle 64Hz 1,917,887 [20] Skoda 10 1 Arms (20 positions) 98Hz ∼ 701, 440 [21]”
A Survey on Human Activity Recognition using Wearable Sensors
Lara, O. D., & Labrador, M. A. (2013) · IEEE Communications Surveys & Tutorials · doi:10.1109/surv.2012.110112.00192 · Cited by 2,662 · cited 1×
“1) Raw inertial data is highly fluctuating over time, making classification impossible when using a single data point in time [15].”
BackgroundA symbolic representation of time series, with implications for streaming algorithms
Lin, J., Keogh, E., Lonardi, S., & Chiu, B. (2003) · Proceedings of the 8th ACM SIGMOD workshop on Research issues in data mining and knowledge discovery · doi:10.1145/882082.882086 · Cited by 1,799 · cited 3×
“Features, such as mean [1], Fourier transforms [2], and symbols [3], are typically extracted from segments of data and then trained using classification methods [4]–[8].”
BackgroundMethodA tutorial on human activity recognition using body-worn inertial sensors
Bulling, A., Blanke, U., & Schiele, B. (2014) · ACM Computing Surveys · doi:10.1145/2499621 · Cited by 1,668 · cited 3×
“Features, such as mean [1], Fourier transforms [2], and symbols [3], are typically extracted from segments of data and then trained using classification methods [4]–[8].”
BackgroundMethod(2015) · Cited by 989
Convolutional Neural Networks for Human Activity Recognition using Mobile Sensors
Zeng, M., Nguyen, L. T., Yu, B., Mengshoel, O. J., Zhu, J., Wu, P., & Zhang, J. (2014) · Proceedings of the 6th International Conference on Mobile Computing, Applications and Services · doi:10.4108/icst.mobicase.2014.257786 · Cited by 876 · cited 2×
“In deep learning, features are abstracted automatically from the data instead of being handcrafted, which allows these machine learning methods to be more effectively used across a range of different classification tasks [10]– [13].”
MethodTime–frequency feature representation using energy concentration: An overview of recent advances
Sejdić, E., Djurović, I., & Jiang, J. (2009) · Digital Signal Processing · doi:10.1016/j.dsp.2007.12.004 · Cited by 756 · cited 1×
“STFT is used to determine the sinusoidal frequency and phase content of local sections of a signal as it changes over time [16].”
MethodWearable Assistant for Parkinson’s Disease Patients With the Freezing of Gait Symptom
Bachlin, M., Plotnik, M., Roggen, D., Maidan, I., Hausdorff, J. M., Giladi, N., & Troster, G. (2010) · IEEE Transactions on Information Technology in Biomedicine · doi:10.1109/titb.2009.2036165 · Cited by 693 · cited 2×
“We obtain comparable accuracy results even against approaches that are more resource demanding [7], [13], [20], [21].”
Result(2006) · doi:10.1007/1-84628-484-8 · Cited by 667
Self-citationA symbolic representation of time series, with implications for streaming algorithms
Lin, J., Keogh, E., Lonardi, S., & Chiu, B. (2003) · doi:10.1145/882085.882086 · Cited by 642
Torch: a modular machine learning software library
Collobert, R., Samy Bengio, A., & Mariéthoz, J. (2002) · Infoscience (Ecole Polytechnique Fédérale de Lausanne) · Cited by 435
A Deep Learning Approach to Human Activity Recognition Based on Single Accelerometer
Chen, Y., & Xue, Y. (2015) · 2015 IEEE International Conference on Systems, Man, and Cybernetics · doi:10.1109/smc.2015.263 · Cited by 378 · cited 2×
“[11] and Chen and Xue [12] both use CNNs with multiple iterations of convolution and subsampling layers, or convolution and pooling layers being applied for feature extraction.”
MethodAnalyzing features for activity recognition
Huynh, T., & Schiele, B. (2005) · Proceedings of the 2005 joint conference on Smart objects and ambient intelligence: innovative context-aware services: usages and technologies · doi:10.1145/1107548.1107591 · Cited by 339
Activity Recognition from On-Body Sensors: Accuracy-Power Trade-Off by Dynamic Sensor Selection
Zappi, P., Lombriser, C., Stiefmeier, T., Farella, E., Roggen, D., Benini, L., & Tröster, G. (2008) · Lecture Notes in Computer Science · doi:10.1007/978-3-540-77690-1_2 · Cited by 329 · cited 2×
“We obtain comparable accuracy results even against approaches that are more resource demanding [7], [13], [20], [21].”
ResultOn the use of ensemble of classifiers for accelerometer-based activity recognition
Catal, C., Tufekci, S., Pirmit, E., & Kocabag, G. (2015) · Applied Soft Computing · doi:10.1016/j.asoc.2015.01.025 · Cited by 240 · cited 1×
“We obtain comparable accuracy results even against approaches that are more resource demanding [7], [13], [20], [21].”
ResultTowards unsupervised physical activity recognition using smartphone accelerometers
Lu, Y., Wei, Y., Liu, L., Zhong, J., Sun, L., & Liu, Y. (2017) · Multimedia Tools and Applications · doi:10.1007/s11042-015-3188-y · Cited by 225 · cited 2×
“Classification methods, such as decision trees, k-Nearest Neighbour (k-NN), and Support Vector Machines (SVM), are then trained to identify different activities using the handcrafted features [4]–[6].”
MethodDeep Activity Recognition Models with Triaxial Accelerometers
Alsheikh, M. A., Selim, A., Niyato, D. T., Doyle, L., Lin, S., & Tan, H.-P. (2015) · DR-NTU (Nanyang Technological University) · Cited by 204
(2014) · Springer eBooks · doi:10.1007/978-1-4471-6374-9 · Cited by 93
Ignatov, A., & Strijov, V. (2015) · Multimedia Tools and Applications · doi:10.1007/s11042-015-2643-0 · Cited by 86
Activity recognition based on accelerometer sensor using combinational classifiers
Zainudin, M. N. S., Sulaiman, M. N., Mustapha, N., & Perumal, T. (2015) · IEEE Conference on Open Systems · doi:10.1109/icos.2015.7377280 · Cited by 46 · cited 2×
“[8], for example, combine decision trees, multilayer perceptron, and logistic regression for HAR.”
MethodReal-time food intake classification and energy expenditure estimation on a mobile device
Ravì, D., Lo, B. P. L., & Yang, G. (2015) · Wearable and Implantable Body Sensor Networks · doi:10.1109/bsn.2015.7299410 · Cited by 38 · cited 2×
“Screenshot of ActiveMiles running on a LG Nexus 5 smartphone – our Android app for recording real world daily human activities [9].”
BackgroundMethodSelf-citation
Research works printed but not linked: 1
- walse, 2016, A study of human activity recognition using AdaBoost classifiers on WISDM dataset, The Institute of Integrative Omics and Applied Biotechnology Journal, 7, 68
Cited by
Deep learning for sensor-based activity recognition: A survey
Wang, J., Chen, Y., Hao, S., Peng, X., & Hu, L. (2019) · Pattern Recognition Letters · doi:10.1016/j.patrec.2018.02.010 · Cited by 2,014
“For multi-modal signals, it is better to use CNN since (Bhattacharya and Lane, 2016) Body-worn, ambiemt RBM Gesture, ADL, transportation Self, D01 (Chen and Xue, 2015) Body-worn CNN ADL Self (Chen et al., 2016b) Body-worn CNN ADL D06 (Edel and K¨oppe, 2016) Body-worn RNN ADL D01, D04, Self (Fang and Hu, 2014) Object,…”
MethodInfluentialDeep Learning for Health Informatics
Ravì, D., Wong, C., Deligianni, F., Berthelot, M., Andreu-Perez, J., Lo, B., & Yang, G.-Z. (2017) · IEEE Journal of Biomedical and Health Informatics · doi:10.1109/jbhi.2016.2636665 · Cited by 2,000
“[96] proposed data preprocessing techniques to standardize and reduce variations in the input data caused by differences in sensor properties, such as placement and orientation.”
MethodSelf-citationA State-of-the-Art Survey on Deep Learning Theory and Architectures
Alom, M. Z., Taha, T. M., Yakopcic, C., Westberg, S., Sidike, P., Nasrin, M. S., Hasan, M., Van Essen, B. C., Awwal, A. A. S., & Asari, V. K. (2019) · Electronics · doi:10.3390/electronics8030292 · Cited by 1,620
Network Intrusion Detection for IoT Security Based on Learning Techniques
Chaabouni, N., Mosbah, M., Zemmari, A., Sauvignac, C., & Faruki, P. (2019) · IEEE Communications Surveys & Tutorials · doi:10.1109/comst.2019.2896380 · Cited by 940
“in [158] where they presented an optimization approach to enable the use of real-time deep learning in low-power devices.”
BackgroundNweke, H. F., Teh, Y. W., Al-garadi, M. A., & Alo, U. R. (2018) · Expert Systems with Applications · doi:10.1016/j.eswa.2018.03.056 · Cited by 930
Metadata
- Authors
- Ravi, D., Wong, C., Lo, B., & Yang, G.-Z.
- Affiliations
- Imperial College London (GB)
- Venue
- 2016 IEEE 13th International Conference on Wearable and Implantable Body Sensor Networks (BSN)
- Publisher
- IEEE
- Published
- 2016-06
- Pages
- 71-76
- ISSN
- 2376-8886 (ISSN-L)
- OpenAlex
- W2495920728
- Cited by
- 269
- References
- 22
- Citations by year
- 2026: 11 · 2025: 22 · 2024: 23 · 2023: 27 · 2022: 44 · 2021: 38 · 2020: 36 · 2019: 30 · 2018: 25 · 2017: 11 · 2016: 2
- FWCI
- 7.87
- Type
- conference paper
- Language
- en
- Open access
- closed
- Locations
- Wearable and Implantable Body Sensor Networks (publishedVersion)
- Spiral (Imperial College London) (submittedVersion)
- Funders
- Engineering and Physical Sciences Research Council (EP/H009744/1, EP/L014149/1, EP/E044298/1)
- Topics
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Context-Aware Activity Recognition Systems
- Physical Sciences › Engineering › Biomedical Engineering › Gait Recognition and Analysis
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Human Pose and Action Recognition
- Keywords
- human activity recognition, inertial data, wearable devices, real-time classification, temporal convolution, resource-efficient deep learning, mobile devices
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.