Human Motion Pattern Recognition Based on the CNN-BiLSTM-Attention Algorithm.

Saved in:

Bibliographic Details
Title:	Human Motion Pattern Recognition Based on the CNN-BiLSTM-Attention Algorithm.
Authors:	Yu, Lie¹ lyu@wtu.edu.cn, Chen, Sijun² 3173578669@qq.com
Source:	IAENG International Journal of Applied Mathematics. Jul2026, Vol. 56 Issue 7, p2509-2523. 15p.
Subjects:	Human activity recognition, Machine learning, Rehabilitation, Medical informatics, Detectors, Sports biomechanics
Abstract:	With increasing healthcare demands, motion pattern recognition analysis has become increasingly critical. This study focuses on level walking, upstairs, downstairs, upslope, and downslope, innovatively proposing a lightweight inertial measurement unit (IMU) sensor motion pattern recognition method based on the convolutional neural network-bidirectional long short-term memory-attention (CNN-BiLSTM-Attention) algorithm. In practical application, ten subjects wore devices equipped with two IMU modules each on the mid-thigh region. These modules precisely captured key motion data--including acceleration and angular velocity--during human movement (3000 sets). Multi-angle discrimination analysis and goodness-of-fit analysis were conducted on the raw data, with the acceleration data along the Y-axis ultimately being selected as the classification source data. Twelve-dimensional feature data were extracted from the experimental data, which were then partitioned into a training set and a test set in a 7:3 ratio. Following multiple comparative experiments, results demonstrate that the CNN-BiLSTM-Attention algorithm outperforms other optimised algorithms. Maximum values of 100% precision, 100% recall, and 99.72% F1 score were achieved across these three metrics, with a peak classification accuracy of 99.44%. With its convenient wearability, reliable data acquisition, and high accuracy, this system provides robust technical support and data assurance for multiple fields, including sports biomechanics research, rehabilitation therapy assessment, and sports training optimisation. [ABSTRACT FROM AUTHOR]
	Copyright of IAENG International Journal of Applied Mathematics is the property of International Association of Engineers (IAENG) and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)
Database:	Engineering Source

Description
Abstract:	With increasing healthcare demands, motion pattern recognition analysis has become increasingly critical. This study focuses on level walking, upstairs, downstairs, upslope, and downslope, innovatively proposing a lightweight inertial measurement unit (IMU) sensor motion pattern recognition method based on the convolutional neural network-bidirectional long short-term memory-attention (CNN-BiLSTM-Attention) algorithm. In practical application, ten subjects wore devices equipped with two IMU modules each on the mid-thigh region. These modules precisely captured key motion data--including acceleration and angular velocity--during human movement (3000 sets). Multi-angle discrimination analysis and goodness-of-fit analysis were conducted on the raw data, with the acceleration data along the Y-axis ultimately being selected as the classification source data. Twelve-dimensional feature data were extracted from the experimental data, which were then partitioned into a training set and a test set in a 7:3 ratio. Following multiple comparative experiments, results demonstrate that the CNN-BiLSTM-Attention algorithm outperforms other optimised algorithms. Maximum values of 100% precision, 100% recall, and 99.72% F1 score were achieved across these three metrics, with a peak classification accuracy of 99.44%. With its convenient wearability, reliable data acquisition, and high accuracy, this system provides robust technical support and data assurance for multiple fields, including sports biomechanics research, rehabilitation therapy assessment, and sports training optimisation. [ABSTRACT FROM AUTHOR]
ISSN:	19929978