Sensors

27 pages, 4945 KB

Open AccessArticle

A Robust Framework for Coffee Bean Package Label Recognition: Integrating Image Enhancement with Vision–Language OCR Models

by Thi-Thu-Huong Le, Yeonjeong Hwang, Ahmada Yusril Kadiptya, JunYoung Son and Howon Kim

Sensors 2025, 25(20), 6484; https://doi.org/10.3390/s25206484 - 20 Oct 2025

Abstract

Text recognition on coffee bean package labels is of great importance for product tracking and brand verification, but it poses a challenge due to variations in image quality, packaging materials, and environmental conditions. In this paper, we propose a pipeline that combines several [...] Read more.

Text recognition on coffee bean package labels is of great importance for product tracking and brand verification, but it poses a challenge due to variations in image quality, packaging materials, and environmental conditions. In this paper, we propose a pipeline that combines several image enhancement techniques and is followed by an Optical Character Recognition (OCR) model based on vision–language (VL) Qwen VL variants, conditioned by structured prompts. To facilitate the evaluation, we construct a coffee bean package image set containing two subsets, namely low-resolution (LRCB) and high-resolution coffee bean image sets (HRCB), enclosing multiple real-world challenges. These cases involve various packaging types (bottles and bags), label sides (front and back), rotation, and different illumination. To address the image quality problem, we design a dedicated preprocessing pipeline for package label situations. We develop and evaluate four Qwen-VL OCR variants with prompt engineering, which are compared against four baselines: DocTR, PaddleOCR, EasyOCR, and Tesseract. Extensive comparison using various metrics, including the Levenshtein distance, Cosine similarity, Jaccard index, Exact Match, BLEU score, and ROUGE scores (ROUGE-1, ROUGE-2, and ROUGE-L), proves significant improvements upon the baselines. In addition, the public POIE dataset validation test proves how well the framework can generalize, thus demonstrating its practicality and reliability for label recognition. Full article

(This article belongs to the Special Issue Digital Imaging Processing, Sensing, and Object Recognition)

► Show Figures

Figure 1

23 pages, 9967 KB

Open AccessArticle

An Optical Water Type-Based Deep Learning Framework for Enhanced Turbidity Estimation in Inland Waters from Sentinel-2 Imagery

by Yue Ma, Qiuyue Chen, Kaishan Song, Qian Yang, Qiang Zheng and Yongchao Ma

Sensors 2025, 25(20), 6483; https://doi.org/10.3390/s25206483 - 20 Oct 2025

Abstract

Turbidity is a crucial and reliable indicator that is extensively utilized in water quality monitoring through remote sensing technology. The development of accurate and applicable models for turbidity estimation is essential. While many existing studies rely on uniform models based on statistical regression [...] Read more.

Turbidity is a crucial and reliable indicator that is extensively utilized in water quality monitoring through remote sensing technology. The development of accurate and applicable models for turbidity estimation is essential. While many existing studies rely on uniform models based on statistical regression or traditional machine learning techniques, the application of deep learning models for turbidity estimation remains limited. This study proposed deep learning models for turbidity estimation based on optical classification of inland waters using Sentinel-2 data. Specifically, the fuzzy c-means (FCM) clustering method was employed to classify optical water types (OWTs) based on their spectral reflectance characteristics. A weighted sum of the turbidity prediction results was generated by the OWT-based convolutional neural network-random forest (CNN-RF) model, with weights derived from the FCM membership degrees. Turbidity for four typical waters was mapped by the proposed method using Sentinel-2 images. The FCM method efficiently classified waters into three OWTs. The OWT-based weighted CNN-RF model demonstrated strong robustness and generalization performance, achieving a high prediction accuracy (R² = 0.900, RMSE = 11.698 NTU). The turbidity maps preserved the spatial continuity of the turbidity distribution and accurately reflected water quality conditions. These findings facilitate the application of deep learning models based on optical classification in turbidity estimation and enhance the capabilities of remote sensing for water quality monitoring. Full article

(This article belongs to the Special Issue Remote Sensing Image Processing, Analysis and Application)

20 pages, 4216 KB

Open AccessArticle

Chaos Fusion Mutation-Based Weighted Mean of Vectors Algorithm for Linear Antenna Array Optimization

by Zhuo Chen, Yan Liu, Liang Dong, Anyong Liu and Yibo Wang

Sensors 2025, 25(20), 6482; https://doi.org/10.3390/s25206482 - 20 Oct 2025

Abstract

This study proposes the Chaos Fusion Mutation-Based Weighted Mean of Vectors Algorithm, an advanced optimization technique within the weighted mean of vectors (INFO) framework for synthesizing unequally spaced linear arrays. The proposed algorithm incorporates three complementary mechanisms: a good-point-set initialization to enhance early [...] Read more.

This study proposes the Chaos Fusion Mutation-Based Weighted Mean of Vectors Algorithm, an advanced optimization technique within the weighted mean of vectors (INFO) framework for synthesizing unequally spaced linear arrays. The proposed algorithm incorporates three complementary mechanisms: a good-point-set initialization to enhance early population coverage, a sine–tent–cosine (STC) chaos–based adaptive parameterization to balance exploration and exploitation, and a normal-cloud mutation to preserve diversity and prevent premature convergence. Array-factor (AF) optimization is posed as a constrained problem, simultaneously minimizing sidelobe level (SLL) and achieving deep-null steering, with penalties applied to enforce geometric and engineering constraints. Across diverse array-synthesis tasks, the proposed algorithm consistently attains lower peak SLLs and more accurate nulls, with faster and more stable convergence than benchmark metaheuristics. Across five simulation scenarios, it demonstrates robust superiority, notably surpassing an enhanced IWO in the combined objectives of deep-null suppression and maximum SLL reduction. In a representative engineering example, we obtain an SLL and a deep null of approximately −32.30 and −125.1 dB, respectively, at 104°. Evaluation of the CEC2020 real-world constrained problems confirms robust convergence and competitive statistical ranking. For reproducibility, all data and code are publicly accessible, as detailed in the Data Availability section. Full article

(This article belongs to the Section Communications)

29 pages, 3862 KB

Open AccessReview

Displacement Self-Sensing Active Magnetic Bearing Drives—An Overview

by Yiling Yang, Yunkai Huang, Fei Peng and Yu Yao

Sensors 2025, 25(20), 6481; https://doi.org/10.3390/s25206481 - 20 Oct 2025

Abstract

Displacement self-sensing active magnetic bearings (AMBs) have garnered significant attention from both academia and industry for their potential to reduce cost, enable system integration, and enhance reliability. While numerous self-sensing methodologies have been researched, the field lacks a unified framework for discussing their [...] Read more.

Displacement self-sensing active magnetic bearings (AMBs) have garnered significant attention from both academia and industry for their potential to reduce cost, enable system integration, and enhance reliability. While numerous self-sensing methodologies have been researched, the field lacks a unified framework for discussing their theoretical foundation and practical applicability. This paper analyzes and summarizes various displacement self-sensing methods, deriving the underlying principles and essence of these techniques, and clarifying the intrinsic interconnections of different schemes. The process of self-sensing is constructed through two steps: online inductance estimation and electromagnetic inductance modeling. A novel framework is then proposed, categorizing online inductance estimation, with dedicated discussion on modeling and handling critical nonlinearity like magnetic saturation and the eddy current effect. Furthermore, this review conducts a systematic comparative analysis, evaluating prevalent schemes against key performance metrics such as robustness, stability, signal-to-noise ratio (SNR), and system complexity. Finally, persistent challenges and future research trends are discussed. This review provides a valuable reference for both researchers and engineers when selecting and implementing self-sensing technologies for AMB systems. Full article

(This article belongs to the Section Physical Sensors)

32 pages, 28980 KB

Open AccessArticle

Probing a CNN–BiLSTM–Attention-Based Approach to Solve Order Remaining Completion Time Prediction in a Manufacturing Workshop

by Wei Chen, Liping Wang, Changchun Liu, Zequn Zhang and Dunbing Tang

Sensors 2025, 25(20), 6480; https://doi.org/10.3390/s25206480 - 20 Oct 2025

Abstract

Manufacturing workshops operate in dynamic and complex environments, where multiple orders are processed simultaneously through interdependent stages. This complexity makes it challenging to accurately predict the remaining completion time of ongoing orders. To address this issue, this paper proposes a data-driven prediction approach [...] Read more.

Manufacturing workshops operate in dynamic and complex environments, where multiple orders are processed simultaneously through interdependent stages. This complexity makes it challenging to accurately predict the remaining completion time of ongoing orders. To address this issue, this paper proposes a data-driven prediction approach that analyzes key features extracted from multi-source manufacturing data. The method involves collecting heterogeneous production data, constructing a comprehensive feature dataset, and applying feature analysis to identify critical influencing factors. Furthermore, a deep learning optimization model based on a Convolutional Neural Network (CNN)–Bidirectional Long Short-Term Memory (BiLSTM)–Attention architecture is designed to handle the temporal and structural complexity of workshop data. The model integrates spatial feature extraction, temporal sequence modeling, and adaptive attention-based refinement to improve prediction accuracy. This unified framework enables the model to learn hierarchical representations, focus on salient temporal features, and deliver accurate and robust predictions. The proposed deep learning predictive model is validated on real production data collected from a discrete manufacturing workshop equipped with typical machines. Comparative experiments with other predictive models demonstrate that the CNN–BiLSTM–Attention model outperforms existing approaches in both accuracy and stability for predicting order remaining completion time, offering strong potential for deployment in intelligent production systems. Full article

(This article belongs to the Section Industrial Sensors)

► Show Figures

Figure 1

15 pages, 3774 KB

Open AccessArticle

MSFDnet: A Multi-Scale Feature Dual-Layer Fusion Model for Sound Event Localization and Detection

by Yi Chen, Zhenyu Huang, Liang Lei and Yu Yuan

Sensors 2025, 25(20), 6479; https://doi.org/10.3390/s25206479 (registering DOI) - 20 Oct 2025

Abstract

The task of Sound Event Localization and Detection (SELD) aims to simultaneously address sound event recognition and spatial localization. However, existing SELD methods face limitations in long-duration dynamic audio scenarios, as they do not fully leverage the complementarity between multi-task features and lack [...] Read more.

The task of Sound Event Localization and Detection (SELD) aims to simultaneously address sound event recognition and spatial localization. However, existing SELD methods face limitations in long-duration dynamic audio scenarios, as they do not fully leverage the complementarity between multi-task features and lack depth in feature extraction, leading to restricted system performance. To address these issues, we propose a novel SELD model—MSDFnet. By introducing a Multi-Scale Feature Aggregation (MSFA) module and a Dual-Layer Feature Fusion strategy (DLFF), MSDFnet captures rich spatial features at multiple scales and establishes a stronger complementary relationship between SED and DOA features, thereby enhancing detection and localization accuracy. On the DCASE2020 Task 3 dataset, our model achieved scores of 0.319, 76%, 10.2°, 82.4%, and 0.198 in

{ER}_{20}

,

F_{20}

,

{LE}_{c d}

,

{LR}_{c d}

, and

{SELD}_{s c o r e}

metrics, respectively. Experimental results demonstrate that MSDFnet performs excellently in complex audio scenarios. Additionally, ablation studies further confirm the effectiveness of the MSFA and DLFF modules in enhancing SELD task performance. Full article

(This article belongs to the Special Issue Sensors and Machine-Learning Based Signal Processing)

► Show Figures

Figure 1

14 pages, 3913 KB

Open AccessArticle

Design Method of a Wide-Field, Dual-Slit, Low-Distortion, and High-Sensitivity Hyperspectral Imager

by Xijie Li, Siyuan Li, Zhinan Zhang, Xiangpeng Feng, Zhong Shen, Xin Lu and Ming Gao

Sensors 2025, 25(20), 6478; https://doi.org/10.3390/s25206478 - 20 Oct 2025

Abstract

To increase target acquisition probability and the signal-to-noise ratio (SNR) of hyperspectral images, this paper presents a wide-field, dual-slit, low-distortion, and high-sensitivity Offner hyperspectral imager, with a wavelength range of 0.4 μm to 0.9 μm, a numerical aperture of 0.15, and a slit [...] Read more.

To increase target acquisition probability and the signal-to-noise ratio (SNR) of hyperspectral images, this paper presents a wide-field, dual-slit, low-distortion, and high-sensitivity Offner hyperspectral imager, with a wavelength range of 0.4 μm to 0.9 μm, a numerical aperture of 0.15, and a slit length of 73 mm. To avoid signal aliasing, the space between the dual slits is 2.4 mm, increasing the SNR by 1.4 times after dual-slit image fusion. Furthermore, to achieve the required registration accuracy of dual-slit images, the spectral performance of the hyperspectral imager is critical. Thus, we compensate and correct the spectral performance and dispersion nonlinearity of the hyperspectral imager by taking advantages of the material properties and tilt eccentricity of a low-dispersion internal reflection curved prism and high-dispersion double-pass curved prisms. To meet the final operation requirements, the tilt of the internal reflection curved prism is used as a compensator. Using the modulation transfer function (MTF) as the evaluation criterion, an inverse sensitivity analysis confirmed that the compensator is a highly sensitive component. Additionally, the root mean square standard deviation (RSS) discrete calculation method was adopted to assess the influence of actual assembly tolerance on spectral performance. The test results demonstrate that the hyperspectral imager meets the registration accuracy requirements of dual-slit images, with an MTF better than 0.4. Furthermore, the spectral smile and spectral keystone of the dual-slit images are both less than or equal to 0.3 pixels. Full article

(This article belongs to the Special Issue Advances in Optical Sensing, Instrumentation and Systems: 2nd Edition)

22 pages, 7167 KB

Open AccessArticle

Ship Ranging Method in Lake Areas Based on Binocular Vision

by Tengwen Zhang, Xin Liu, Mingzhi Shao, Yuhan Sun and Qingfa Zhang

Sensors 2025, 25(20), 6477; https://doi.org/10.3390/s25206477 - 20 Oct 2025

Abstract

The unique hollowed-out catamaran hulls and complex environmental conditions in lake areas hinder traditional ranging algorithms (combining target detection and stereo matching) from accurately obtaining depth information near the center of ships. This not only impairs the navigation of electric tourist boats but [...] Read more.

The unique hollowed-out catamaran hulls and complex environmental conditions in lake areas hinder traditional ranging algorithms (combining target detection and stereo matching) from accurately obtaining depth information near the center of ships. This not only impairs the navigation of electric tourist boats but also leads to high computing resource consumption. To address this issue, this study proposes a ranging method integrating improved ORB (Oriented FAST and Rotated BRIEF) with stereo vision technology. Combined with traditional optimization techniques, the proposed method calculates target distance and angle based on the triangulation principle, providing a rough alternative solution for the “gap period” of stereo matching-based ranging. The method proceeds as follows: first, it acquires ORB feature points with relatively uniform global distribution from preprocessed binocular images via a local feature weighting approach; second, it further refines feature points within the ROI (Region of Interest) using a quadtree structure; third, it enhances matching accuracy by integrating the FLANN (Fast Library for Approximate Nearest Neighbors) and PROSAC (Progressive Sample Consensus) algorithms; finally, it applies the screened matching point pairs to the triangulation method to obtain the position and distance of the target ship. Experimental results show that the proposed algorithm improves processing speed by 6.5% compared with the ORB-PROSAC algorithm. Under ideal conditions, the ranging errors at 10m and 20m are 2.25% and 5.56%, respectively. This method can partially compensate for the shortcomings of stereo matching in ranging under the specified lake area scenario. Full article

(This article belongs to the Section Sensing and Imaging)

23 pages, 1492 KB

Open AccessArticle

Probabilistic Sampling Networks for Hybrid Structure Planning in Semi-Structured Environments

by Xiancheng Ji, Jianjun Yi and Lin Su

Sensors 2025, 25(20), 6476; https://doi.org/10.3390/s25206476 - 20 Oct 2025

Abstract

The advancement of adaptable industrial robots in intelligent manufacturing is hindered by the inefficiency of traditional motion planning methods in high-dimensional spaces. Therefore, a Dempster–Shafer evidence theory-based hybrid motion planner is proposed, in which a probabilistic sampling network (PSNet) and an enhanced artificial [...] Read more.

The advancement of adaptable industrial robots in intelligent manufacturing is hindered by the inefficiency of traditional motion planning methods in high-dimensional spaces. Therefore, a Dempster–Shafer evidence theory-based hybrid motion planner is proposed, in which a probabilistic sampling network (PSNet) and an enhanced artificial potential field (EAPF) cooperate with each other to improve the planning performance. The PSNet architecture comprises two modules: a motion planning module (MPM) and a fusion sampling module (FSM). The MPM utilizes sensor data alongside the robot’s current and target configurations to recursively generate diverse multimodal distributions of the next configuration. Based on the distribution information, the FSM was used as a decision-maker to ultimately generate globally connectable paths. Moreover, the FSM is equipped to correct collision path points caused by network inaccuracies through Gaussian resampling. Simultaneously, an augmented artificial potential field with a dynamic rotational field is deployed to repair local paths when worst-case collision scenarios occur. This collaborative strategy harmoniously unites the complementary strengths of both components, thereby enhancing the overall resilience and adaptability of the motion planning system. Experiments were conducted in various environments. The results demonstrate that the proposed method can quickly find directly connectable paths in diverse environments while reliably avoiding sudden obstacles. Full article

(This article belongs to the Special Issue Advanced Robotic Manipulators and Control Applications)

23 pages, 6273 KB

Open AccessArticle

Photovoltaic Module Degradation Detection Using V–P Curve Derivatives and LSTM-Based Classification

by Chan-Ho Lee, Sang-Kil Lim, Sung-Jun Park and Beom-Hun Kim

Sensors 2025, 25(20), 6475; https://doi.org/10.3390/s25206475 - 20 Oct 2025

Abstract

Photovoltaic systems are a core component of eco-friendly energy technologies and are now widely utilized across the world for power generation. However, solar modules that are continuously exposed to the external environment experience gradual performance degradation, which results in significant power loss and [...] Read more.

Photovoltaic systems are a core component of eco-friendly energy technologies and are now widely utilized across the world for power generation. However, solar modules that are continuously exposed to the external environment experience gradual performance degradation, which results in significant power loss and operational problems. Existing aging diagnostic methods such as current–voltage curve analysis and electroluminescence/photoluminescence testing have limitations in terms of real-time monitoring, quantitative evaluation, and applicability to large-scale power plants. To address these challenges, this study proposes a novel degradation detection method that utilizes the first-order derivative of the voltage–power curve of solar modules to extract key features. This method can estimate the number of degraded solar modules within a string and the degree of degradation, enabling early detection of subtle changes in electrical characteristics. In this study, we developed an AI model based on long short-term memory to classify normal and abnormal states and predict aging status, thereby supporting monitoring and early diagnosis. The model architecture was designed to reflect the characteristics of solar power systems, adopting a relatively shallow network due to the time-series data not being excessively long and the feature changes being clear. This design effectively mitigates the issues of overfitting and gradient vanishing, thereby positively contributing to the stability of model training. The training and validation results of the proposed long short-term memory model were verified through MATLAB simulations, confirming its effectiveness in learning and convergence. Full article

(This article belongs to the Special Issue Condition Monitoring of Electrical Equipment Within Power Systems)

21 pages, 3498 KB

Open AccessArticle

Effect of Walking Speed on the Reliability of a Smartphone-Based Markerless Gait Analysis System

by Edilson Fernando de Borba, Jorge L. Storniolo, Serena Cerfoglio, Paolo Capodaglio, Veronica Cimolin, Leonardo A. Peyré-Tartaruga, Marcus P. Tartaruga and Paolo Cavallari

Sensors 2025, 25(20), 6474; https://doi.org/10.3390/s25206474 - 20 Oct 2025

Abstract

Quantitative gait analysis is essential for understanding motor function and guiding clinical decisions. While marker-based motion capture (MoCap) systems are accurate, they are costly and require specialized facilities. OpenCap, a markerless alternative, offers a more accessible approach; however, its reliability across different walking [...] Read more.

Quantitative gait analysis is essential for understanding motor function and guiding clinical decisions. While marker-based motion capture (MoCap) systems are accurate, they are costly and require specialized facilities. OpenCap, a markerless alternative, offers a more accessible approach; however, its reliability across different walking speeds remains uncertain. This study assessed the agreement between OpenCap and MoCap in measuring spatiotemporal parameters, joint kinematics, and center of mass (CoM) displacement during level walking at three speeds: slow, self-selected, and fast. Fifteen healthy adults performed multiple trials simultaneously, recorded by both systems. Agreement was analyzed using intraclass correlation coefficients (ICC), minimal detectable change (MDC), Bland–Altman analyses, root mean square error (RMSE), Statistical Parametric Mapping (SPM), and repeated-measures ANOVA. Results indicated excellent agreement for spatiotemporal variables (ICC ≥ 0.95) and high consistency for joint waveforms (RMSE < 2°) and CoM displacement (RMSE < 6 mm) across all speeds. However, the joint range of motion (ROM) showed lower reliability, especially at the hip and ankle, at higher speeds. ANOVA revealed no significant System × Speed interactions for most variables, though a significant effect of speed was noted, with OpenCap underestimating walking speed more at fast speeds. Overall, OpenCap is a valuable tool for gait assessment, very accurate for spatiotemporal data and CoM displacement. Still, caution should be taken when interpreting joint kinematics and speed at different walking speeds. Full article

(This article belongs to the Special Issue Sensors and Data Analysis for Biomechanics and Physical Activity)

► Show Figures

Figure 1

32 pages, 3278 KB

Open AccessReview

Advancing Circular Economy Implementation for High-Speed Train Rolling Stocks by the Integration of Digital Twins and Artificial Intelligence

by Lalitphat Khongsomchit and Sakdirat Kaewunruen

Sensors 2025, 25(20), 6473; https://doi.org/10.3390/s25206473 - 20 Oct 2025

Abstract

This paper presents a state-of-the-art review on the integration of digital twins and artificial intelligence to advance the circular economy and the 10R principles implementation in high-speed train rolling stock. Rolling stock generates substantial waste at the end of its service life, yet [...] Read more.

This paper presents a state-of-the-art review on the integration of digital twins and artificial intelligence to advance the circular economy and the 10R principles implementation in high-speed train rolling stock. Rolling stock generates substantial waste at the end of its service life, yet the application of the circular economy and the 10R principles (Refuse, Rethink, Reduce, Reuse, Repair, Refurbish, Remanufacture, Repurpose, Recycle, and Recover) in this domain remains limited compared with infrastructure. The review analyses 47 studies retrieved from Web of Science and IEEE Xplore, focusing on digital twin applications in railway infrastructure and rolling stock, and machine learning techniques. Findings reveal that most studies concentrate on data management and efficiency improvement, while only a few explicitly address the circular economy and 10R principles. A comparative analysis of high-waste components against current machine learning applications further highlights critical gaps. To address these, an automated workflow is proposed, incorporating digital twins, artificial intelligence, and the 10R principles to support condition monitoring and sustainable resource management. The study provides insights and research directions to enhance sustainability in railway asset management. Full article

(This article belongs to the Special Issue Advanced Sensing Technologies for Sustainable and Resilient Railway Infrastructures)

► Show Figures

Figure 1

28 pages, 1892 KB

Open AccessReview

Wearable Devices in Healthcare Beyond the One-Size-Fits All Paradigm

by Elena Giovanna Bignami, Anna Fornaciari, Sara Fedele, Mattia Madeo, Matteo Panizzi, Francesco Marconi, Erika Cerdelli and Valentina Bellini

Sensors 2025, 25(20), 6472; https://doi.org/10.3390/s25206472 - 20 Oct 2025

Abstract

Wearable devices (WDs) are increasingly integrated into clinical workflows to enable continuous, non-invasive vital signs monitoring. Combined with Artificial Intelligence (AI), these systems can shift clinical monitoring from being reactive to predictive, allowing for earlier detection of deterioration and more personalized interventions. The [...] Read more.

Wearable devices (WDs) are increasingly integrated into clinical workflows to enable continuous, non-invasive vital signs monitoring. Combined with Artificial Intelligence (AI), these systems can shift clinical monitoring from being reactive to predictive, allowing for earlier detection of deterioration and more personalized interventions. The value of these technologies lies not in absolute measurements, but in detecting physiological parameters trends relative to each patient’s baseline. Such a trend-based approach enables real-time prediction of deterioration, enhancing patient safety and continuity of care. However, despite their shared multiparametric capabilities, WDs are not interchangeable. This narrative review analyzes nine clinically validated devices, Radius VSM^® (Masimo Corporation, Irvine, CA, USA), BioButton^® (BioIntelliSense Inc., Redwood City, CA, USA. Distributed by Medtronic), Portrait Mobile^® (GE HealthCare, Chicago, IL, USA), VitalPatch^® (VitalConnect Inc., San Jose, CA, USA), CardioWatch 287-2^® (Corsano Health B.V., The Hague, The Netherlands. Distributed by Medtronic), Cosinuss C-Med Alpha^® (Cosinuss Gmb, Munich, Germany), SensiumVitals^® (Sensium Healthcare Limited, Abingdon, Oxfordshire, UK), Isansys Lifetouch^® (Isansys Lifecare Ltd., Abingdon, Oxfordshire, UK), and CheckPoint Cardio^® (CheckPoint R&D LTD., Kazanlak, Bulgaria), highlighting how differences in sensor configurations, battery life, connectivity, and validation contexts influence their suitability across various clinical environments. Rather than establishing a hierarchy of technical superiority, this review emphasizes the importance of context-driven selection, considering care setting, patient profile, infrastructure requirements, and interoperability. Each device demonstrates strengths and limitations depending on patient population and operational demands, ranging from perioperative, post-operative, emergency, or post-Intensive Care Unit (ICU) settings. The findings support a tailored approach to WD implementation, where matching device capabilities to clinical needs is key to maximizing utility, safety, and efficiency. Full article

(This article belongs to the Section Wearables)

► Show Figures

Figure 1

25 pages, 2621 KB

Open AccessArticle

Analysis of a Driving Simulator’s Steering System for the Evaluation of Autonomous Vehicle Driving

by Juan F. Dols, Samuel Boix, Jaime Molina, Sara Moll, Francisco J. Camacho and Griselda López

Sensors 2025, 25(20), 6471; https://doi.org/10.3390/s25206471 - 20 Oct 2025

Abstract

The integration of autonomous vehicles (AVs) into road transport requires robust experimental tools to analyze the human–machine interaction, particularly under conditions of system disengagement. This study presents the primary controls calibration and virtual scenario validation of the EVACH autonomous driving simulator, designed to [...] Read more.

The integration of autonomous vehicles (AVs) into road transport requires robust experimental tools to analyze the human–machine interaction, particularly under conditions of system disengagement. This study presents the primary controls calibration and virtual scenario validation of the EVACH autonomous driving simulator, designed to reproduce the SAE Level 2 and Level 3 driving modes in rural road scenarios. The simulator was customized through hardware and software developments including a dedicated data acquisition system to ensure the accurate detection of braking, steering, and other critical control inputs. Calibration tests demonstrated high fidelity, with minor errors in brake and steering control measurements, consistent with values observed in production vehicles. To validate the virtual driving rural environment, comparative experiments were conducted between naturalistic road tests and simulator-based autonomous driving, where five volunteers participated in the preliminary pilot test. Results showed that average speeds in the simulation closely matched those recorded on real roads, with differences of less than 1 km/h with minimum standard deviation and confidence values. These findings confirm that the EVACH simulator provides a stable and faithful reproduction of autonomous driving conditions. The experimental platform offers valuable support for current and future research on the safe deployment of automated vehicles. Full article

(This article belongs to the Special Issue Multi-Sensor Fusion for Autonomous Vehicles and Driver Assistance Systems)

► Show Figures

Figure 1

17 pages, 2479 KB

Open AccessArticle

A Semi-Automatic Labeling Framework for PCB Defects via Deep Embeddings and Density-Aware Clustering

by Sang-Jeong Lee, Sung-Bal Seo and You-Suk Bae

Sensors 2025, 25(20), 6470; https://doi.org/10.3390/s25206470 - 19 Oct 2025

Abstract

(1) Background. Printed circuit board (PCB) inspection is increasingly constrained by the cost and latency of reliable labels, owing to tiny/low-contrast defects embedded in complex backgrounds and severe class imbalance. (2) Methods. We proposed a semi-automatic labeling pipeline that converts anomaly detection proposals [...] Read more.

(1) Background. Printed circuit board (PCB) inspection is increasingly constrained by the cost and latency of reliable labels, owing to tiny/low-contrast defects embedded in complex backgrounds and severe class imbalance. (2) Methods. We proposed a semi-automatic labeling pipeline that converts anomaly detection proposals into class labels via small margin cropping from images, interchangeable embeddings (HOG, ResNet-50, ViT-B/16), clustering (k-means/GMM/HDBSCAN), and cluster-level verification using representative montages. (3) Results. On 9354 cropped defects spanning 10 categories (imbalance IR ≈ 1542, Gini ≈ 0.642), ResNet-50 + HDBSCAN achieved NMI ≈ 0.290, AMI ≈ 0.283, and purity ≈ 0.624 with ~47 clusters; ViT + HDBSCAN was comparable (NMI ≈ 0.281, AMI ≈ 0.274, ~44 clusters). With a fixed taxonomy, k-means (K = 10) yielded the strongest ARI (0.169 with ResNet-50; 0.158 with ViT). Macro-purity exceeded micro-purity, indicating many small, homogeneous clusters suitable for one-shot acceptance/rejection, enabling an upper-bound ~200× reduction in operator decisions relative to per-image labeling. (4) Conclusions. The workflow provides an auditable, resource-flexible path from normal-only localization to scalable supervision, prioritizing labeling productivity over detector state-of-the-art and directly addressing the industrial bottleneck in the development lifecycle for PCB inspection. Full article

(This article belongs to the Special Issue Computer Vision and Sensing Technologies for Industrial Quality Inspection: 2nd Edition)

► Show Figures

Figure 1

24 pages, 5023 KB

Open AccessArticle

Internet of Things Node with Real-Time LoRa GEO Satellite Connectivity for Agrifood Chain Tracking in Remote Areas

by Giacomo Giannetti, Marco Badii, Giovanni Lasagni, Stefano Maddio, Giovanni Collodi, Monica Righini and Alessandro Cidronali

Sensors 2025, 25(20), 6469; https://doi.org/10.3390/s25206469 - 19 Oct 2025

Abstract

This work presents an Internet of Things (IoT) node designed for low-power agrifood chain tracking in remote areas, where long-range terrestrial communication is either unavailable or severely limited. The novelty of this study lies in the development and characterization of an IoT node [...] Read more.

This work presents an Internet of Things (IoT) node designed for low-power agrifood chain tracking in remote areas, where long-range terrestrial communication is either unavailable or severely limited. The novelty of this study lies in the development and characterization of an IoT node prototype that leverages direct-to-satellite connectivity through a geostationary Earth orbit (GEO) satellite, using long-range frequency-hopping spread spectrum (LR-FHSS) modulation in the licensed S-band. The prototype integrates a microcontroller unit that manages both the radio modem and a suite of sensors, enclosed in a plastic box suitable for field deployment. Characterization in an anechoic chamber demonstrated a maximum effective isotropic radiated power (EIRP) of

27.5

dBm, sufficient to establish a reliable satellite link. The onboard sensors provide global positioning as well as measurements of acceleration, temperature, humidity, and solar radiation intensity. Prototype performance was assessed in two representative scenarios: stationary and mobile. Regarding energy consumption, the average charge drained by the radio modem per transmission cycle was measured to be 356 mC. With a battery pack composed of four 2500 mAh NiMH cells, the estimated upper bound on the number of transmitted packets is approximately 25,000. Full article

(This article belongs to the Special Issue Enabling IoT Sensors Through Satellite Networks: Applications, Technologies and Solutions)

23 pages, 4804 KB

Open AccessArticle

Particle Image Velocimetry Algorithm Based on Spike Camera Adaptive Integration

by Xiaoqiang Li, Changxu Wu, Yichao Wang, Hongyuan Li, Yuan Li, Tiejun Huang, Yuhao Huang and Pengyu Lv

Sensors 2025, 25(20), 6468; https://doi.org/10.3390/s25206468 - 19 Oct 2025

Abstract

In particle image velocimetry (PIV), overexposure is particularly common in regions with high illumination. In particular, strong scattering or background reflection at the liquid–gas interface will make the overexposure phenomenon more obvious, resulting in local pixel saturation, which will significantly reduce the particle [...] Read more.

In particle image velocimetry (PIV), overexposure is particularly common in regions with high illumination. In particular, strong scattering or background reflection at the liquid–gas interface will make the overexposure phenomenon more obvious, resulting in local pixel saturation, which will significantly reduce the particle image quality, and thus reduce the particle recognition rate and the accuracy of velocity field estimation. This study addresses the overexposure challenges in particle image velocimetry applications, mainly to address the challenge that the velocity field cannot be measured due to the difficulty in effectively detecting particles in the exposed area. In order to address the challenge of overexposure, this paper does not use traditional frame-based high-speed cameras, but instead proposes a particle image velocimetry algorithm based on adaptive integral spike camera data using a neuromorphic vision sensor (NVS). Specifically, by performing target-background segmentation on high-frequency digital spike signals, the method suppresses high illumination background regions and thus effectively mitigates overexposure. Then the spike data are further adaptively integrated based on both regional background illumination characteristics and the spike frequency features of particles with varying velocities, resulting in high signal-to-noise ratio (SNR) reconstructed particle images. Flow field computation is subsequently conducted using the reconstructed particle images, with validation through both simulation and experiment. In simulation, in the overexposed area, the average flow velocity estimation error of frame-based cameras is 8.594 times that of spike-based cameras. In the experiments, the spike camera successfully captured continuous high-density particle trajectories, yielding measurable and continuous velocity fields. Experimental results demonstrate that the proposed particle image velocimetry algorithm based on the adaptive integration of the spike camera effectively addresses overexposure challenges caused by high illumination of the liquid–gas interface in flow field measurements. Full article

(This article belongs to the Special Issue Real-Time Object Detection and Classification Using Advanced Sensing Techniques)

► Show Figures

Figure 1

17 pages, 2003 KB

Open AccessArticle

Performance Assessment of Multistatic/Multi-Frequency 3D GPR Imaging by Linear Microwave Tomography

by Mehdi Masoodi, Gianluca Gennarelli, Carlo Noviello, Ilaria Catapano and Francesco Soldovieri

Sensors 2025, 25(20), 6467; https://doi.org/10.3390/s25206467 - 19 Oct 2025

Abstract

The advent of multichannel ground-penetrating radar systems capable of acquiring multiview, multistatic, and multifrequency data is offering new possibilities to improve subsurface imaging performance. However, this raises the need for reconstruction approaches capable of handling such sophisticated configurations and the resulting increase in [...] Read more.

The advent of multichannel ground-penetrating radar systems capable of acquiring multiview, multistatic, and multifrequency data is offering new possibilities to improve subsurface imaging performance. However, this raises the need for reconstruction approaches capable of handling such sophisticated configurations and the resulting increase in the data volume. Therefore, the challenge lies in identifying proper measurement configurations that balance image quality with the complexity and duration of data acquisition. As a contribution to this topic, the present paper focuses on a measurement system working in reflection mode and composed of an array of antennas, consisting of a transmitting antenna and several receiving antennas, whose spatial offset is comparable to the probing wavelength. Therefore, for each position of the transmitting antenna, a single-view/multistatic configuration is considered. The imaging task is solved by adopting a linear microwave tomographic approach, which provides a qualitative reconstruction of the investigated scenario. In particular, a 3D inverse scattering problem is tackled for an isotropic, homogeneous, lossless, and non-magnetic medium under the Born approximation, considering both single- and multi-frequency data. A preliminary analysis, referring to a 3D free-space reference scenario, is performed in terms of the spectral content of the scattering operator and the system’s point spread function. Finally, an experimental validation under laboratory conditions is presented in order to verify the expected imaging capability of the inversion approach. Full article

(This article belongs to the Special Issue Radars, Sensors and Applications for Applied Geophysics)

► Show Figures

Figure 1

26 pages, 3460 KB

Open AccessArticle

Classification and Clustering of Fiber Break Events in Thermoset CFRP Using Acoustic Emission and Machine Learning

by Richard Dela Amevorku, David Amoateng-Mensah, Manoj Rijal and Mannur J. Sundaresan

Sensors 2025, 25(20), 6466; https://doi.org/10.3390/s25206466 - 19 Oct 2025

Abstract

Carbon Fiber-Reinforced Polymer (CFRP) composites, widely used across industries, exhibit various damage mechanisms depending on the loading conditions applied. This study employs a structural health monitoring (SHM) approach to investigate the three primary failure modes, fiber breakage, matrix cracking, and delamination, in thermoset [...] Read more.

Carbon Fiber-Reinforced Polymer (CFRP) composites, widely used across industries, exhibit various damage mechanisms depending on the loading conditions applied. This study employs a structural health monitoring (SHM) approach to investigate the three primary failure modes, fiber breakage, matrix cracking, and delamination, in thermoset quasi-isotropic CFRPs subjected to quasi-static tensile loading until failure. Acoustic emission (AE) signals acquired from an experiment were leveraged to analyze and classify these real-time signals into the failure modes using machine learning (ML) techniques. Due to the extensive number of AE signals recorded during testing, manually classifying these failure mechanisms through waveform inspection was impractical. ML, alongside ensemble learning, algorithms were implemented to streamline the classification, making it more efficient, accurate, and reliable. Conventional AE parameters from the data acquisition system and feature extraction techniques applied to the recorded waveforms were implemented exclusively as classification features to investigate their reliability and accuracy in classifying failure modes in CFRPs. The classification models exhibited up to 99% accuracy, as depicted by evaluation metrics. Further studies, using cross-correlation techniques, ascertained the presence of fiber break events occurring in the bundles as the thermoset CFRP composite approached failure. These findings highlight the significance of integrating machine learning into SHM for the early detection of real-time damage and effective monitoring of residual life in composite materials. Full article

(This article belongs to the Special Issue Sensing and Machine Learning Control: Progress and Applications)

► Show Figures

Figure 1

Journal Description

Sensors

Latest Articles

Journal Menu

Journal Browser

Highly Accessed Articles

Latest Books

E-Mail Alert

News

Topics

Conferences

Special Issues

Topical Collections

Further Information

Guidelines

MDPI Initiatives

Follow MDPI