Cited by Research on an Elderly Indoor Fall Detection System Based on IoT and Computer Vision Technology: Integration and Performance Analysis of YOLOv5 and OpenPose

Research on an Elderly Indoor Fall Detection System Based on IoT and Computer Vision Technology: Integration and Performance Analysis of YOLOv5 and OpenPose DOI

Highlights in Science Engineering and Technology, Journal Year: 2024, Volume and Issue: 119, P. 415 - 419

Published: Dec. 11, 2024

The rapid growth of the global elderly population has led to an increased need for efficient and reliable fall detection systems ensure timely medical assistance. Traditional monitoring methods, such as wearable devices environmental sensors, often face challenges in accuracy, reliability, user compliance. This study proposes indoor system elderly, integrating Internet Things (IoT) computer vision technologies, specifically utilizing YOLOv5 OpenPose algorithms. is employed accurate human body detection, providing position information by computing center objects frames or videos. then used detailed real-time pose estimation, detecting 135 key points on body—including hands, face, feet—to assess whether a occurred based posture analysis. was tested using two datasets, GMDCSA URFD, achieving sensitivity values 0.9412 0.9583, respectively, indicating high accuracy detection. false positive rates were 0.0588 0.2857, while negative 0.0417, demonstrating system's reliability minimizing errors. integration leverages their combined strengths object resulting robust solution suitable applications.

Language: Английский

A self-powered wearable sensor for infant fall detection based on triboelectric nanogenerator DOI

Luoke Hu,

Hui Meng,

Zhonggui Xu

et al.

Applied Physics A, Journal Year: 2025, Volume and Issue: 131(3)

Published: Feb. 4, 2025

Language: Английский

Citations

RGANet: A Human Activity Recognition Model for Extracting Temporal and Spatial Features from WiFi Channel State Information DOI

Jianyuan Hu, Fei Ge, Xinyu Cao

et al.

Sensors, Journal Year: 2025, Volume and Issue: 25(3), P. 918 - 918

Published: Feb. 3, 2025

With the rapid advancement of communication technologies, wireless networks have not only transformed people’s lifestyles but also spurred development numerous emerging applications and services. Against this backdrop, research on Wi-Fi-based human activity recognition (HAR) has become a hot topic in both academia industry. Channel State Information (CSI) contains rich spatiotemporal information. However, existing deep learning methods for typically focus either temporal or spatial features. While some approaches do combine types features, they often emphasize sequences underutilize In contrast, paper proposes an enhanced approach by modifying residual (ResNet) instead using simple CNN. This modification allows effective feature extraction while preserving The extracted features are then fed into GRU model sequence learning. Our achieves accuracy 99.4% UT_HAR dataset 99.24% NTU-FI HAR dataset. Compared to other models, RGANet shows improvements 1.21% 0.38%

Language: Английский

Citations

SSARS: Secure smart-home activity recognition system DOI

C. Anna Palagan,

Tilak Raj,

N. Muthuvairavan Pillai

et al.

Computers & Electrical Engineering, Journal Year: 2025, Volume and Issue: 123, P. 110203 - 110203

Published: Feb. 28, 2025

Language: Английский

Citations

Enhanced human activity recognition in medical emergencies using a hybrid deep CNN and bi-directional LSTM model with wearable sensors DOI

Nishanth Adithya Chandramouli,

Sivaramakrishnan Natarajan, Amal H. Alharbi

et al.

Scientific Reports, Journal Year: 2024, Volume and Issue: 14(1)

Published: Dec. 28, 2024

Human activity recognition (HAR) is one of the most important segments technology advancement in applications smart devices, healthcare systems & fitness. HAR uses details from wearable sensors that capture way human beings move or engage with their surrounding. Several researchers have thus presented different ways modeling motion, and some been as follows: Many methods movements. Therefore, this paper, we proposed CNN BiLSTM model undersampling to improve actions. The evaluated using state-of-the-art metrics, including accuracy, precision, recall, F1-score, on two publicly available datasets: For instance, MHEALTH Actitracker. This will enable team attain test accuracies up 98.5% dataset. CNN-BiLSTM outperforms conventional deep learning methods, reported Actitracker dataset, by about 5% improvement. has many applications, which used keep vigil over elderly people who live alone alert when fallen any strange movement noticed could be a sign individual experiencing medical Emergency. It can also applied physiotherapy, where patient's development throughout rehabilitation exercises accessed.

Language: Английский

Citations

Vision-based Human Fall Detection Systems: A Review DOI