Human Jumping Knee-Bending Angle Recognition Model Based on Multi-Channel Attention

DING Lei; FANG Chen’an; HU Xinrong; LI Yang; MING Wenkai; LIN Lin; WU Yuan

doi:10.11907/rjdk.251093

您当前的位置：

首页 >

文章列表页 >

Human Jumping Knee-Bending Angle Recognition Model Based on Multi-Channel Attention

Artificial Intelligence | 更新时间：2026-04-30

- Human Jumping Knee-Bending Angle Recognition Model Based on Multi-Channel Attention
- Software Guide Vol. 25, Issue 4, Pages: 48-56(2026)
- 作者机构：
  
  1.武汉纺织大学计算机与人工智能学院
  2.武汉纺织大学体育部，湖北武汉430200
- 作者简介：
- 基金信息：
- DOI：10.11907/rjdk.251093
  CLC： TP391.41
- Received：18 March 2025，
  
  Published：15 April 2026
- 稿件说明：
移动端阅览
丁磊,方晨安,胡新荣等.基于多通道注意力的人体跳跃屈膝角度识别模型[J].软件导刊,2026,25(04):48-56.

DING Lei,FANG Chen’an,HU Xinrong,et al.Human Jumping Knee-Bending Angle Recognition Model Based on Multi-Channel Attention[J].Software Guide,2026,25(04):48-56.
丁磊,方晨安,胡新荣等.基于多通道注意力的人体跳跃屈膝角度识别模型[J].软件导刊,2026,25(04):48-56. DOI： 10.11907/rjdk.251093.

DING Lei,FANG Chen’an,HU Xinrong,et al.Human Jumping Knee-Bending Angle Recognition Model Based on Multi-Channel Attention[J].Software Guide,2026,25(04):48-56. DOI： 10.11907/rjdk.251093.

摘要

目前，深度学习与可穿戴设备的结合在运动定量评估领域得到了广泛应用，并展现出重要的理论价值和应用潜力。然而，传统神经网络方法在处理具有不同运动质量的细粒度动作时，难以有效提取关键的多维度特征，从而限制了其在精确识别和深入分析个体运动行为时的性能。为了解决这一问题，提出一种多通道注意力模型（Conv+CEA+AG），旨在通过多通道注意力（CEA）机制增强对动作重要特征的捕捉能力，从而提升运动行为分析的精确性。首先，采集不同屈膝角度下的人体跳跃运动数据，并对传感器信号进行预处理，提取有效的跳跃周期信号以构建跳跃运动数据集。其次，将集成多通道注意力（CEA）的神经网络应用于数据集，构建了跳跃运动屈膝角度检测模型。实验结果表明，该模型在分类任务中达到了94.32%的准确率，在回归任务中实现了2.04°的平均绝对误差。与传统的神经网络模型（Conv+GRU）及现有模型（Conv+CIE+AG）相比，准确率分别提升了12.47%和6.08%，平均绝对误差分别下降了3.49°和3.09°，证明了该模型在屈膝角度识别方面表现更优异。

Abstract

Currently， the combination of deep learning and wearable devices is widely used in the field of quantitative movement assessment and shows important theoretical value and application potential. However， traditional neural network methods are difficult to effectively extract key multidimensional features when dealing with fine-grained actions with different motion qualities， which limits their performance in accurately recognizing and deeply analyzing individual motion behaviors. To address this problem， a multi-channel attention model（Conv+CEA+AG） is proposed， aiming to enhance the ability to capture important features of the action through the multi-channel attention （CEA） mechanism， and thus improve the accuracy of the analysis of motor behavior. Firstly， human jumping motion data under different knee flexion angles are collected， and the sensor signals are preprocessed to extract effective jumping cycle signals in order to construct a jumping motion dataset. Subsequently， a neural network with integrated multi-channel attention is applied to the dataset to construct a jumping motion knee-bending angle detection model. The experimental results show that the proposed model achieves 94.32% accuracy in the classification task and 2.04° mean absolute error in the regression task. Compared with the traditional neural network model （Conv+GRU） and the previous proposed model （Conv+CIE+AG）， the accuracy is improved by 12.47% and 6.08%， and the average absolute error decrease by 3.49° and 3.09°， respectively， which prove that the model demonstrate a better performance in knee-bending angle recognition.

关键词

Keywords

references

ZHANG Y W ， LI R X ， DAI J W ， et al . Activity recognition algorithm based on multi-feature fusion ［J］. Software Guide ， 2022 ， 21 （ 3 ）： 72 - 77 .

张耀威，李瑞祥，戴健威，等 . 基于多特征融合的活动识别算法［J］. 软件导刊， 2022 ， 21 （ 3 ）： 72 - 77 .

DRAZAN J F ， PHILLIPS W T ， SEETHAPATHI N ， et al . Moving outside the lab： markerless motion capture accurately quantifies sagittal plane kinematics during the vertical jump ［J］. Journal of Biomechanics ， 2021 ， 125 ： 110547 . doi: 10.1101/2021.03.16.435503 http://dx.doi.org/10.1101/2021.03.16.435503

ARULAMPALAM J ， PIERREPONT J ， KARK L . Markerless motion capture： validity of Microsoft Kinect cameras and iPisoft ［J］. Gait & Posture ， 2015 ， 42 ： S76 . doi: 10.1016/j.gaitpost.2015.06.141 http://dx.doi.org/10.1016/j.gaitpost.2015.06.141

MASCIA G ， DE LAZZARI B ， CAMOMILLA V . Machine learning aided jump height estimate democratization through smartphone measures ［J］. Frontiers in Sports and Active Living ， 2023 ， 5 ： 1112739 . doi: 10.3389/fspor.2023.1112739 http://dx.doi.org/10.3389/fspor.2023.1112739

LU Y Z ， WANG H ， HU F ， et al . Effective recognition of human lower limb jump locomotion phases based on multi-sensor information fusion and machine learning ［J］. Medical & Biological Engineering & Computing ， 2021 ， 59 ： 883 - 899 . doi: 10.1007/s11517-021-02335-9 http://dx.doi.org/10.1007/s11517-021-02335-9

JAITNER T ， SCHMIDT M ， NOLTE K ， et al . Vertical jump diagnosis for multiple athletes using a wearable inertial sensor unit ［J］. Sports Technology ， 2015 ， 8 （ 1 ）： 51 - 57 . doi: 10.1080/19346182.2015.1117476 http://dx.doi.org/10.1080/19346182.2015.1117476

SUN T ， LI D X ， FAN B F ， et al . Real-time ground reaction force and knee extension moment estimation during drop landings via modular LSTM modeling and wearable IMUs ［J］. IEEE Journal of Biomedical and Health Informatics ， 2023 ， 27 （ 7 ）： 3222 - 3233 . doi: 10.1109/JBHI.2023.3268239 http://dx.doi.org/10.1109/JBHI.2023.3268239

OLAWORE I . Measurement of ankle joint kinematics using IMUs during countermovement jumps and lateral skater jumps ［J］. Open Access Library Journal ， 2024 ， 11 （ 7 ）： 1 - 13 . doi: 10.4236/oalib.1111809 http://dx.doi.org/10.4236/oalib.1111809

ABEDIN A ， EHSANPOUR M ， SHI Q ， et al . Attend and discriminate： beyond the state-of-the-art for human activity recognition using wearable sensors ［J］. Proceedings of the ACM on Interactive， Mobile， Wearable and Ubiquitous Technologies ， 2021 ， 5 （ 1 ）： 1 - 22 . doi: 10.1145/3448083 http://dx.doi.org/10.1145/3448083

WOCHATZ M ， TILGNER N ， MUELLER S ， et al . Reliability and validity of the Kinect V2 for the assessment of lower extremity rehabilitation exercises ［J］. Gait & Posture ， 2019 ， 70 ： 330 - 335 . doi: 10.1016/j.gaitpost.2019.03.020 http://dx.doi.org/10.1016/j.gaitpost.2019.03.020

FREITAS-JUNIOR C G D ， PAES P P ， FORTES L D S ， et al . Reliability of the high-speed camera-based system （HSC-Kinovea） for lower-limb explosive strength endurance assessment in athletes ［J］. Journal of Physical Education ， 2021 ， 32 （ 1 ）： 1 - 8 . doi: 10.4025/jphyseduc.v32i1.3220 http://dx.doi.org/10.4025/jphyseduc.v32i1.3220

STRUTZENBERGER G ， KANKO R ， SELBIE S ， et al . Assessment of kinematic CMJ data using a deep learning algorithm-based markerless motion capture system ［J］. ISBS Proceedings Archive ， 2021 ， 39 （ 1 ）： 236 .

QUAGLIARELLA L ， SASANELLI N ， BELGIOVINE G ， et al . Evaluation of standing vertical jump by ankles acceleration measurement ［J］. The Journal of Strength & Conditioning Research ， 2010 ， 24 （ 5 ）： 1229 - 1236 . doi: 10.1519/jsc.0b013e3181cb281a http://dx.doi.org/10.1519/jsc.0b013e3181cb281a

GRAINGER M ， WEISBERG A ， STERGIOU P ， et al . Comparison of two methods in the estimation of vertical jump height ［J］. Journal of Human Sport and Exercise ， 2020 ， 15 （ 3 ）： 623 - 632 . doi: 10.14198/jhse.2020.153.12 http://dx.doi.org/10.14198/jhse.2020.153.12

ORDÓÑEZ F J ， ROGGEN D . Deep convolutional and LSTM recurrent neural networks for multimodal wearable activity recognition ［J］. Sensors ， 2016 ， 16 （ 1 ）： 115 . doi: 10.3390/s16010115 http://dx.doi.org/10.3390/s16010115

HOU Q B ， ZHOU D Q ， FENG J S . Coordinate attention for efficient mobile network design ［C］// Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition ， 2021 ： 13713 - 13722 . doi: 10.1109/cvpr46437.2021.01350 http://dx.doi.org/10.1109/cvpr46437.2021.01350

GUO M H ， LIU Z N ， MU T J ， et al . Beyond self-attention： external attention using two linear layers for visual tasks ［J］. IEEE Transactions on Pattern Analysis and Machine Intelligence ， 2022 ， 45 （ 5 ）： 5436 - 5447 .

GUO M H ， CAI J X ， LIU Z N ， et al . PCT： point cloud transformer ［J］. Computational Visual Media ， 2021 ， 7 ： 187 - 199 . doi: 10.1007/s41095-021-0229-5 http://dx.doi.org/10.1007/s41095-021-0229-5

MURAHARI V S ， PLÖTZ T . On attention models for human activity recognition ［C］// Proceedings of the 2018 ACM International Symposium on Wearable Computers ， 2018 ： 100 - 103 . doi: 10.1145/3267242.3267287 http://dx.doi.org/10.1145/3267242.3267287

WANG J ， ZHENG V W ， CHEN Y ， et al . Deep transfer learning for cross-domain activity recognition［DB/OL］ . https：//arxiv.org/pdf/1807.07963 https://arxiv.org/pdf/1807.07963 . doi: 10.1145/3265689.3265705 http://dx.doi.org/10.1145/3265689.3265705

WEN Y D ， ZHANG K P ， LI Z F ， et al . A discriminative feature learning approach for deep face recognition ［C］// European Conference on Computer Vision ， 2016 ： 499 - 515 . doi: 10.1007/978-3-319-46478-7_31 http://dx.doi.org/10.1007/978-3-319-46478-7_31

ZHANG Q L ， YANG Y B . SA-Net： shuffle attention for deep convolutional neural networks ［C］// 2021 IEEE International Conference on Acoustics， Speech and Signal Processing ， 2021 ： 2235 - 2239 . doi: 10.1109/icassp39728.2021.9414568 http://dx.doi.org/10.1109/icassp39728.2021.9414568

MISRA D ， NALAMADA T ， ARASANIPALAI A U ， et al . Rotate to attend： convolutional triplet attention module ［C］// Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision ， 2021 ： 3139 - 3148 . doi: 10.1109/wacv48630.2021.00318 http://dx.doi.org/10.1109/wacv48630.2021.00318

LI X ， WANG W H ， HU X L ， et al . Selective kernel networks ［C］// Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition ， 2019 ： 510 - 519 . doi: 10.1109/cvpr.2019.00060 http://dx.doi.org/10.1109/cvpr.2019.00060

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Sensitive Text Recognition Based on Deep Bidirectional LSTM and MultiHeadAttention

A Review of Deep Learning-Based Adrenal CT Image Segmentation Research

Innovative Applications and Improvements of Deep Learning in the Detection and Recognition of Oracle-bone Character Images

Research on Prediction Method of Soil Available Nitrogen Content in Guangxi Based on SA-CNN-LSTM

Remote Sensing Image Object Detection Method Integrating Segmentation and Classification Models

Related Author

ZHU Dingju

DENG Renfeng

WU Chao

FANG Ziyang

WANG Jinliang

XIAO Peiyao

LI Yikai

ZHANG Zhi

Related Institution

School of Artificial Intelligence， South China Normal University

School of Health Science and Engineering，University of Shanghai for Science and Technology

Department of Cardiology，Shanghai General Hospital，Shanghai Jiao Tong University School of Medicine

School of Computer Science， China University of Geosciences

School of Computer Science and Engineering， Guilin University of Technology

⁰