東華大學圖書館 |

語系: 繁體中文

說明(常見問題)

回圖書館首頁

手機版館藏查詢

登入

回首頁

切換: 標籤 | MARC模式 | ISBD

FindBook

Google Book

Amazon

博客來

Artificial Intelligence-based Approaches for Automated Detection of Irregular Walking Surfaces with Wearable Sensor.

紀錄類型:	書目-電子資源 : Monograph/item
正題名/作者:	Artificial Intelligence-based Approaches for Automated Detection of Irregular Walking Surfaces with Wearable Sensor./
作者:	Ng, Hui Ru.
出版者:	Ann Arbor : ProQuest Dissertations & Theses, : 2022,
面頁冊數:	89 p.
附註:	Source: Masters Abstracts International, Volume: 84-01.
Contained By:	Masters Abstracts International84-01.
標題:	Computer science. -
電子資源:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=29163327
ISBN:	9798834000082

Artificial Intelligence-based Approaches for Automated Detection of Irregular Walking Surfaces with Wearable Sensor.
Ng, Hui Ru.

Artificial Intelligence-based Approaches for Automated Detection of Irregular Walking Surfaces with Wearable Sensor. - Ann Arbor : ProQuest Dissertations & Theses, 2022 - 89 p.

Source: Masters Abstracts International, Volume: 84-01.

Thesis (M.S.)--University of Nebraska at Omaha, 2022.

This item must not be sold to any third party vendors.

The walkability of a neighborhood is associated with public health, economic and environmental benefits. Sidewalk walking surface condition is a significant indicator of walkability as it supports and encourages pedestrian travel and physical activity. However, common sidewalk assessment practices are subjective, inefficient, and ineffective. Alternate methods proposed in previous studies for objective and automated assessment of sidewalk surfaces using urban data do not consider pedestrians' physiological responses. Furthermore, no studies have proposed methods to detect irregular walking surfaces that incorporate gait analysis utilizing a wearable sensor. Therefore, to contribute towards developing an objective, real-time monitoring system that considers human bodily responses, the goals of this study are to identify the most suitable location for sensor placement and to explore the feasibility of using machine learning and deep learning approaches for the purpose of classifying good and irregular walking surfaces with a single accelerometer.In this study, we conducted experiments on 12 subjects with sensors attached to three different locations and collected walking data on good and irregular walking surfaces. We found that the most suitable location for sensor placement for detecting irregular walking surfaces is on the ankle. Utilizing acceleration data from the right ankle, we extracted gait features and trained five machine learning classifiers for our machine learning approach and Long Short-Term Memory (LSTM) neural networks for our deep learning approach. In our machine learning approach, we found that Support Vector Machine (SVM) was the best model because it was the most generalizable subject-wise and was able to achieve an average Area Under the receiver operating characteristics Curve (AUC) of 80% when evaluated using leave-one-subject-out as the test set protocol. As for the deep learning approach, the LSTM model trained with gait features was found to achieve more superior results compared to the LSTM model trained with raw acceleration data. The classification performance for LSTM and SVM was further improved with post-processing. These results demonstrated that the SVM or LSTM model trained with accelerometer-based gait features can be used as an objective tool for sidewalk surface condition assessment.

ISBN: 9798834000082Subjects--Topical Terms:

523869
Computer science.
Subjects--Index Terms:

Deep learning

Artificial Intelligence-based Approaches for Automated Detection of Irregular Walking Surfaces with Wearable Sensor.
LDR:03677nmm a2200433 4500 001 2349187
005 20220920135800.5
008 241004s2022 ||||||||||||||||| ||eng d
020 $a 9798834000082
035 $a (MiAaPQ)AAI29163327
035 $a AAI29163327
040 $a MiAaPQ $c MiAaPQ
100 1 $a Ng, Hui Ru. $3 3688587
245 1 0 $a Artificial Intelligence-based Approaches for Automated Detection of Irregular Walking Surfaces with Wearable Sensor.
260 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2022
300 $a 89 p.
500 $a Source: Masters Abstracts International, Volume: 84-01.
500 $a Advisor: Youn, Jong-Hoon.
502 $a Thesis (M.S.)--University of Nebraska at Omaha, 2022.
506 $a This item must not be sold to any third party vendors.
520 $a The walkability of a neighborhood is associated with public health, economic and environmental benefits. Sidewalk walking surface condition is a significant indicator of walkability as it supports and encourages pedestrian travel and physical activity. However, common sidewalk assessment practices are subjective, inefficient, and ineffective. Alternate methods proposed in previous studies for objective and automated assessment of sidewalk surfaces using urban data do not consider pedestrians' physiological responses. Furthermore, no studies have proposed methods to detect irregular walking surfaces that incorporate gait analysis utilizing a wearable sensor. Therefore, to contribute towards developing an objective, real-time monitoring system that considers human bodily responses, the goals of this study are to identify the most suitable location for sensor placement and to explore the feasibility of using machine learning and deep learning approaches for the purpose of classifying good and irregular walking surfaces with a single accelerometer.In this study, we conducted experiments on 12 subjects with sensors attached to three different locations and collected walking data on good and irregular walking surfaces. We found that the most suitable location for sensor placement for detecting irregular walking surfaces is on the ankle. Utilizing acceleration data from the right ankle, we extracted gait features and trained five machine learning classifiers for our machine learning approach and Long Short-Term Memory (LSTM) neural networks for our deep learning approach. In our machine learning approach, we found that Support Vector Machine (SVM) was the best model because it was the most generalizable subject-wise and was able to achieve an average Area Under the receiver operating characteristics Curve (AUC) of 80% when evaluated using leave-one-subject-out as the test set protocol. As for the deep learning approach, the LSTM model trained with gait features was found to achieve more superior results compared to the LSTM model trained with raw acceleration data. The classification performance for LSTM and SVM was further improved with post-processing. These results demonstrated that the SVM or LSTM model trained with accelerometer-based gait features can be used as an objective tool for sidewalk surface condition assessment.
590 $a School code: 1060.
650 4 $a Computer science. $3 523869
650 4 $a Physiology. $3 518431
650 4 $a Information technology. $3 532993
650 4 $a Artificial intelligence. $3 516317
650 4 $a Kinesiology. $3 517627
650 4 $a Information science. $3 554358
653 $a Deep learning
653 $a Gait analysis
653 $a Machine learning
653 $a Sidewalk surface assessment
653 $a Walkability
653 $a Wearable sensor
653 $a Automated detection
690 $a 0984
690 $a 0489
690 $a 0575
690 $a 0800
690 $a 0719
690 $a 0723
710 2 $a University of Nebraska at Omaha. $b Computer Science. $3 1020791
773 0 $t Masters Abstracts International $g 84-01.
790 $a 1060
791 $a M.S.
792 $a 2022
793 $a English
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=29163327