A Novel Framework Based on Deep Learning Architecture for Continuous Human Activity Recognition with Inertial Sensors

Sensors, Год журнала: 2024, Номер 24(7), С. 2199 - 2199

Опубликована: Март 29, 2024

Frameworks for human activity recognition (HAR) can be applied in the clinical environment monitoring patients’ motor and functional abilities either remotely or within a rehabilitation program. Deep Learning (DL) models exploited to perform HAR by means of raw data, thus avoiding time-demanding feature engineering operations. Most works targeting with DL-based architectures have tested workflow performance on data related separate execution tasks. Hence, paucity literature has been found regard frameworks aimed at recognizing continuously executed actions. In this article, authors present design, development, testing continuous (CHAR). The model was trained recorded from ten healthy subjects eight different subjects. Despite limited sample size, claim capability proposed framework accurately classify actions feasible time, making it potentially useful scenario.

Язык: Английский

---

Adaptive Feedback-Driven Segmentation for Continuous Multi-Label Human Activity Recognition

Applied Sciences, Год журнала: 2025, Номер 15(6), С. 2905 - 2905

Опубликована: Март 7, 2025

Radar-based continuous human activity recognition (HAR) in realistic scenarios faces challenges segmenting and classifying overlapping or concurrent activities. This paper introduces a feedback-driven adaptive segmentation framework for multi-label classification HAR, leveraging Bayesian optimization (BO) reinforcement learning (RL) to dynamically adjust parameters such as segment length overlap the data stream, optimizing them based on performance metrics accuracy F1-score. Using public dataset of activities, method trains ResNet18 models spectrogram, range-Doppler, range-time representations from 20% computational subset. Then, it scales optimized full dataset. Comparative analysis against fixed-segmentation baselines was made. The results demonstrate significant improvements performance, confirming potential techniques enhancing efficiency HAR systems.

Язык: Английский

---

Transfer Learning using Computer Vision Models for Fall Detection from UWB Radars

Transportation research procedia, Год журнала: 2025, Номер 84, С. 19 - 26

Опубликована: Янв. 1, 2025

Язык: Английский

---

Radar-Based Elderly Fall Detection Using Smoothed Pseudo Wigner Ville Distribution and XGBoost Learning

Studies in health technology and informatics, Год журнала: 2024, Номер unknown

Опубликована: Авг. 22, 2024

Falls among the elderly population pose significant health risks, often leading to morbidity and decreased quality of life. Traditional fall detection methods, namely wearable devices cameras, have limitations such as lighting conditions privacy concerns. Radar-based has emerged a promising alternative, offering unobtrusive technique. In this study, an attempt been made classify using smoothed pseudo wigner-ville distribution (SPWVD) images XGBoost learning. For this, online publicly available radar database (N=15) is considered. Radar signals employed SPWVD for time-frequency representation images. Ten features are extracted applied Experiments performed performance evaluated 10-fold cross validation. The proposed approach able discriminate fall. Using learning, yields maximum average classification accuracy, f1-score, precision, sensitivity, specificity, kappa scores 87.47%, 87.38%, 88.12%, 86.81%, 88.31% 74.94% respectively. combination conventional with concentration measures median frequency obtained second best performance. Thus, framework could be utilized accurate efficient falls in their private spaces.

Язык: Английский

---

A hybrid vision transformer and residual neural network model for fall detection using UWB radars

Applied Intelligence, Год журнала: 2024, Номер 55(3)

Опубликована: Дек. 26, 2024

Язык: Английский

---

Human Activity Recognition and Classification Systems using YOLOv9

Опубликована: Окт. 25, 2024

Язык: Английский

---