Poster: Motion Sensor Based Dragging Feet Detection Using Lightweight Classification Model

Dragging feet is an important indicator of potential falls. We propose using a motion sensor-embedded wearable device deployed near the ankle to sense walking motions and detect dragging feet continuously. To adapt to the limited computation ability and power supply of the wearable device, only seve...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:IEEE/ACM Conference on Connected Health: Applications, Systems and Engineering Technologies (Online) S. 191 - 192
Hauptverfasser: Zhou, Michelle, Wang, Shuangquan
Format: Tagungsbericht
Sprache:Englisch
Veröffentlicht: IEEE 19.06.2024
Schlagworte:
Accuracy
Adaptation models
Computational modeling
Costs
dragging feet detection
Feature extraction
Legged locomotion
lightweight model
motion sensor
Power supplies
ISSN:2832-2975
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:Dragging feet is an important indicator of potential falls. We propose using a motion sensor-embedded wearable device deployed near the ankle to sense walking motions and detect dragging feet continuously. To adapt to the limited computation ability and power supply of the wearable device, only several most representative features are selected to build a lightweight classification model. This model can achieve high recognition accuracy while reducing computation and energy costs. Experimental results show that using only five features common to all users, the proposed method can accurately distinguish normal walking from dragging feet. The average accuracy of the four algorithms is greater than or equal to 93.0% for all five users and up to 98.7% for user 2.
ISSN:2832-2975
DOI:10.1109/CHASE60773.2024.00034