针对传统人体跌倒检测在复杂姿态与环境下的精度和误检难题,提出改进 YOLOv5s算法。引入多尺度卷积注意力机制（MSCA）,通过并行多尺度卷积核与通道混合强化多尺度特征感知;优化EIoU损失函数,增加姿态长宽比约束、高斯加权中心点回归及尺度自适应惩罚,提升定位精度。在FDD数据集上,改进模型精确率91.5%、召回率91.6%,mAP50和mAP50-95达93.7%和75.9%,较基线分别提升5.5%、7.5%、6.0%和5.4%,显著增强复杂场景的鲁棒性,为人体智能监护提供有效方案。
Abstract
The improved YOLOv5s algorithm is proposed to address the accuracy and false detection challenges of traditional human fall detection in complex poses and environments. Multi-scale Convolutional Attention Mechanism (MSCA) is introduced to strengthen multi-scale feature sensing by parallel multi-scale convolutional kernel and channel mixing; the EIoU loss function is optimized, and pose aspect ratio constraints, Gaussian-weighted centroid regression, and scale-adaptive penalties are added to improve the positioning accuracy. On the FDD dataset, the improved model precision rate is 91.5%, the recall rate is 91.6%, and the mAP50 and mAP50-95 reach 93.7% and 75.9%, which are 5.5%, 7.5%, 6.0%, and 5.4% higher than the baseline, respectively, which significantly enhances the robustness of the complex scene, and provides an effective solution for human body intelligent monitoring.
关键词
跌倒检测 /
YOLOv5s /
多尺度卷积注意力机制 /
EIoU损失函数 /
长宽比约束
Key words
fall detection /
YOLOv5s /
multi-scale convolutional attention mechanism /
EIoU loss function /
aspect ratio constraint
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] GBD 2019 Ageing Collaborators.Global,regional,and national burden of diseases and injuries for adults 70 years and older: systematic analysis for the Global Burden of Disease 2019 StudyBD 2019 Ageing Collaborators.Global,regional,and national burden of diseases and injuries for adults 70 years and older: systematic analysis for the Global Burden of Disease 2019 Study[J].BMJ,2022:376.
[2] 王冬子,郭政鑫,桂林卿,等.基于环境振动的IR-UWB超感知范围人体跌倒检测方法[J].计算机研究与发展, 2024, 62(11):1-18.
[3] 冯梓文,冯云霞.基于改进YOLOv5的人体跌倒检测算法[J].电子设计工程,2025,33(1):1-6.
[4] 张帆,姜文刚.基于改进的YOLOv3算法的船舶目标检测[J].计算机与数字工程,2024,52(5):1348-1352.
[5] Yang Z,Wang X,Li J.EIoU:An improved vehicle detection algorithm based on vehiclenet neural network[C]//Journal of physics:Conference series.IOP Publishing,2021,1924(1):012001.
[6] Wang Y,Liang C,Wang X,et al.MSCA-YOLO:A YOLOv5-based Steel Defect Detection Method Enhanced with Multi-Scale Feature Extraction and Contextual Augmentation[J].Journal of Imaging Science & Technology,2024,68(4):1-10.
[7] Kepski M,Kwolek B.Fall detection using ceiling-mounted 3d depth camera[C]//2014 International conference on computer vision theory and applications (VISAPP).IEEE,2014,2:640-647.
[8] Rougier C,Meunier J,St-Arnaud A,et al.Robust video surveillance for fall detection based on human shape deformation[J].IEEE Transactions on circuits and systems for video Technology,2011,21(5):611-622.
[9] Wang X,Jia K.Human fall detection algorithm based on YOLOv3[C]//2020 IEEE 5th International Conference on Image, Vision and Computing (ICIVC).IEEE,2020:50-54.
[10] Kan X,Zhu S,Zhang Y,et al.A lightweight human fall detection network[J].Sensors,2023,23(22):9069.
[11] Ren S,He K,Girshick R,et al.Faster R-CNN:Towards real-time object detection with region proposal networks[J].IEEE transactions on pattern analysis and machine intelligence,2016,39(6):1137-1149.
[12] Hou Q,Zhou D,Feng J.Coordinate attention for efficient mobile network design[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition.2021:13713-13722.
[13] Zhu X,Cheng D,Zhang Z,et al.An empirical study of spatial attention mechanisms in deep networks[C]//Proceedings of the IEEE/CVF international conference on computer vision.2019:6688-6697.
[14] Bui T,Liu J,Cao J,et al.Elderly fall detection in complex environment based on improved YOLOv5s and LSTM[J].Applied Sciences,2024,14(19):9028.
[15] Rezatofighi H,Tsoi N,Gwak J Y,et al.Generalized intersection over union:A metric and a loss for bounding box regression[C]//Proceedings of the IEEE/CVF conference on computer vision and pattern recognition.2019:658-666.
[16] Zheng Z,Wang P,Liu W,et al.Distance-IoU loss:Faster and better learning for bounding box regression[C]//Proceedings of the AAAI conference on artificial intelligence.2020,34(07):12993-13000.
[17] Zhang Y,Ren W,Zhang Z,et al.Focal and Efficient IOU Loss for Accurate Bounding Box Regression[J].Neurocomputing, 2022,506:146-157.
基金
福建省中青年教师教育科研项目（科技类）“基于 2-DCNN 的人体异常姿态检测系统”,项目编号：JAT220483
