Fusion from Multimodal Gait Spatiotemporal Data for Human Gait Speed Classifications

Abdullah Alharthi; Krikor Ozanyan

doi:10.1109/SENSORS47087.2021.9639816

Fusion from Multimodal Gait Spatiotemporal Data for Human Gait Speed Classifications

Research output: Contribution to conference › Paper › peer-review

Abstract

Human gait pattens remain largely undefined when relying on a single sensing modality. We report a pilot implementation of sensor fusion to classify gait spatiotemporal signals, from a publicly available dataset of 50 participants, harvested from four different type of sensors. For fusion we propose a hybrid Convolutional Neural Network and Long Short- Term Memory (hybrid CNN+LSTM) and Multi-stream CNN. The classification results are compared to single modality data using Single-stream CNN, a state-of-the-art Vision Transformer, and statistical classifiers algorithms. The fusion models outperformed the single modality methods and classified gait speed of previously unseen 10 random subjects with 97% F1-score prediction accuracy of the four gait speed classes.

Original language	English
Pages	1
Number of pages	4
DOIs	https://doi.org/10.1109/SENSORS47087.2021.9639816
Publication status	Published - 31 Oct 2021
Event	IEEE SENSORS 2021 - Virtual conference, Sydney, Australia Duration: 31 Oct 2021 → 4 Nov 2021 https://2021.ieee-sensorsconference.org/

Conference

Conference	IEEE SENSORS 2021
Country/Territory	Australia
City	Sydney
Period	31/10/21 → 4/11/21
Internet address	https://2021.ieee-sensorsconference.org/

Keywords

Deep Convolutional Neural Networks (CNN)
Gait Speed
Long Short-Term Memory (LSTM)
Multimodal Data
Transformers

Access to Document

10.1109/SENSORS47087.2021.9639816Licence: Unspecified

Cite this

@conference{bd28e38c8a8340d9993eed1741d60404,

title = "Fusion from Multimodal Gait Spatiotemporal Data for Human Gait Speed Classifications",

abstract = "Human gait pattens remain largely undefined when relying on a single sensing modality. We report a pilot implementation of sensor fusion to classify gait spatiotemporal signals, from a publicly available dataset of 50 participants, harvested from four different type of sensors. For fusion we propose a hybrid Convolutional Neural Network and Long Short- Term Memory (hybrid CNN+LSTM) and Multi-stream CNN. The classification results are compared to single modality data using Single-stream CNN, a state-of-the-art Vision Transformer, and statistical classifiers algorithms. The fusion models outperformed the single modality methods and classified gait speed of previously unseen 10 random subjects with 97% F1-score prediction accuracy of the four gait speed classes.",

keywords = "Deep Convolutional Neural Networks (CNN), Gait Speed, Long Short-Term Memory (LSTM), Multimodal Data, Transformers",

author = "Abdullah Alharthi and Krikor Ozanyan",

note = "Publisher Copyright: {\textcopyright} 2021 IEEE.; IEEE SENSORS 2021 ; Conference date: 31-10-2021 Through 04-11-2021",

year = "2021",

month = oct,

day = "31",

doi = "10.1109/SENSORS47087.2021.9639816",

language = "English",

pages = "1",

url = "https://2021.ieee-sensorsconference.org/",

}

TY - CONF

T1 - Fusion from Multimodal Gait Spatiotemporal Data for Human Gait Speed Classifications

AU - Alharthi, Abdullah

AU - Ozanyan, Krikor

PY - 2021/10/31

Y1 - 2021/10/31

N2 - Human gait pattens remain largely undefined when relying on a single sensing modality. We report a pilot implementation of sensor fusion to classify gait spatiotemporal signals, from a publicly available dataset of 50 participants, harvested from four different type of sensors. For fusion we propose a hybrid Convolutional Neural Network and Long Short- Term Memory (hybrid CNN+LSTM) and Multi-stream CNN. The classification results are compared to single modality data using Single-stream CNN, a state-of-the-art Vision Transformer, and statistical classifiers algorithms. The fusion models outperformed the single modality methods and classified gait speed of previously unseen 10 random subjects with 97% F1-score prediction accuracy of the four gait speed classes.

AB - Human gait pattens remain largely undefined when relying on a single sensing modality. We report a pilot implementation of sensor fusion to classify gait spatiotemporal signals, from a publicly available dataset of 50 participants, harvested from four different type of sensors. For fusion we propose a hybrid Convolutional Neural Network and Long Short- Term Memory (hybrid CNN+LSTM) and Multi-stream CNN. The classification results are compared to single modality data using Single-stream CNN, a state-of-the-art Vision Transformer, and statistical classifiers algorithms. The fusion models outperformed the single modality methods and classified gait speed of previously unseen 10 random subjects with 97% F1-score prediction accuracy of the four gait speed classes.

KW - Deep Convolutional Neural Networks (CNN)

KW - Gait Speed

KW - Long Short-Term Memory (LSTM)

KW - Multimodal Data

KW - Transformers

UR - http://www.scopus.com/inward/record.url?scp=85123643285&partnerID=8YFLogxK

U2 - 10.1109/SENSORS47087.2021.9639816

DO - 10.1109/SENSORS47087.2021.9639816

M3 - Paper

SP - 1

T2 - IEEE SENSORS 2021

Y2 - 31 October 2021 through 4 November 2021

ER -

Fusion from Multimodal Gait Spatiotemporal Data for Human Gait Speed Classifications

Abstract

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this