Sensors

52 pages, 17474 KiB

Open AccessFeature PaperArticle

LSEWOA: An Enhanced Whale Optimization Algorithm with Multi-Strategy for Numerical and Engineering Design Optimization Problems

by Junhao Wei, Yanzhao Gu, Yuzheng Yan, Zikun Li, Baili Lu, Shirou Pan and Ngai Cheong

Sensors 2025, 25(7), 2054; https://doi.org/10.3390/s25072054 (registering DOI) - 25 Mar 2025

Abstract

The Whale Optimization Algorithm (WOA) is a bio-inspired metaheuristic algorithm known for its simple structure and ease of implementation. However, WOA suffers from issues such as premature convergence, low population diversity in the later stages of iteration, slow convergence rate, low convergence accuracy, [...] Read more.

The Whale Optimization Algorithm (WOA) is a bio-inspired metaheuristic algorithm known for its simple structure and ease of implementation. However, WOA suffers from issues such as premature convergence, low population diversity in the later stages of iteration, slow convergence rate, low convergence accuracy, and an imbalance between exploration and exploitation. In this paper, we proposed an enhanced whale optimization algorithm with multi-strategy (LSEWOA). LSEWOA employs Good Nodes Set Initialization to generate uniformly distributed whale individuals, a newly designed Leader-Followers Search-for-Prey Strategy, a Spiral-based Encircling Prey strategy inspired by the concept of Spiral flight, and an Enhanced Spiral Updating Strategy. Additionally, we redesigned the update mechanism for convergence factor a to better balance exploration and exploitation. The effectiveness of the proposed LSEWOA was evaluated using CEC2005, and the impact of each improvement strategy was analyzed. We also performed a quantitative analysis of LSEWOA and compare it with other state-of-art metaheuristic algorithms in 30/50/100 dimensions. Finally, we applied LSEWOA to nine engineering design optimization problems to verify its capability in solving real-world optimization challenges. Experimental results demonstrate that LSEWOA outperformed better than other algorithms and successfully addressed the shortcomings of the classic WOA. Full article

(This article belongs to the Section Intelligent Sensors)

► Show Figures

Figure 1

18 pages, 3146 KiB

Open AccessArticle

Multi-Objective Optimization Method for High-Efficiency and Low-Consumption Wire Rope Greasing Process

by Fan Zhou, Yuemin Wang, Ruqing Gong and Binghui Tang

Sensors 2025, 25(7), 2053; https://doi.org/10.3390/s25072053 - 25 Mar 2025

Abstract

Wire rope greasing is essential for protecting wire ropes from corrosion and wear. To address issues such as low maintenance efficiency and excessive grease usage, this study proposes a high-efficiency, low-consumption optimization control method for the wire rope greasing process. A time objective [...] Read more.

Wire rope greasing is essential for protecting wire ropes from corrosion and wear. To address issues such as low maintenance efficiency and excessive grease usage, this study proposes a high-efficiency, low-consumption optimization control method for the wire rope greasing process. A time objective function for the greasing process and a consumption objective function for grease are established. Considering the actual constraints of greasing equipment performance and greasing quality, a multi-objective optimization model is developed with greasing speed, greasing thickness, grease flow rate, and greasing time as the optimization parameters. The model aims to achieve high efficiency (minimizing greasing process time) and low consumption (minimizing grease consumption). Weight coefficients are introduced to transform the multi-objective optimization model into a single-objective optimization model, which is then solved using an improved genetic algorithm. The effectiveness of the model is validated through a specific case study, and a sensitivity analysis of the weight coefficients of the objective functions in the optimization model is conducted. This research provides valuable support for wire rope greasing process planning and improvement. Full article

(This article belongs to the Section Intelligent Sensors)

24 pages, 1549 KiB

Open AccessArticle

Situational Awareness Prediction for Remote Tower Controllers Based on Eye-Tracking and Heart Rate Variability Data

by Weijun Pan, Ruihan Liang, Yuhao Wang, Dajiang Song and Zirui Yin

Sensors 2025, 25(7), 2052; https://doi.org/10.3390/s25072052 - 25 Mar 2025

Abstract

Remote tower technology is an important development direction for air traffic control to reduce the construction and operation costs of small or remote airports. However, its digital and virtualized working environment poses new challenges to controllers’ situational awareness (SA). In this study, a [...] Read more.

Remote tower technology is an important development direction for air traffic control to reduce the construction and operation costs of small or remote airports. However, its digital and virtualized working environment poses new challenges to controllers’ situational awareness (SA). In this study, a dataset is constructed by collecting eye-tracking (ET) and heart rate variability (HRV) data from participants in a remote tower simulation control experiment. At the same time, probe questions are designed that correspond to the SA hierarchy in conjunction with the remote tower control task flow, and the dataset is annotated using the scenario presentation assessment method (SPAM). The annotated dataset containing 25 ET and HRV features is trained using the LightGBM model optimized by a Tree-structured Parzen Estimator, and feature selection and model interpretation are performed using the SHapley Additive exPlanations (SHAP) analysis. The results show that the TPE-LightGBM model exhibits excellent prediction capability, obtaining an RMSE, MAE and adjusted R² of 0.0909, 0.0730 and 0.7845, respectively. This study presents an effective method for assessing and predicting controllers’ SA in remote tower environments. It further provides a theoretical basis for understanding the effect of the physiological state of remote tower controllers on their SA. Full article

(This article belongs to the Section Biosensors)

12 pages, 911 KiB

Open AccessArticle

Light-Controlled Soft Switches for Optical Logic Gate Operations

by Chuang Wang, Hao Wu, Quanwang Niu, Xiaohong Yan and Xiangfu Wang

Sensors 2025, 25(7), 2051; https://doi.org/10.3390/s25072051 - 25 Mar 2025

Abstract

While liquid crystal elastomers (LCEs) show promise for diverse soft actuators due to their strong stimulus responsiveness, limited investigation into their light perception and processing restricts their wider use in intelligent systems. This study employs a hollow double-layer structure to design light-controlled logic [...] Read more.

While liquid crystal elastomers (LCEs) show promise for diverse soft actuators due to their strong stimulus responsiveness, limited investigation into their light perception and processing restricts their wider use in intelligent systems. This study employs a hollow double-layer structure to design light-controlled logic soft switches based on LCEs. The design realizes digital logic circuits including AND gates, OR gates, and NOT gates, as well as an optical switch array capable of converting light signals into visualized digital signals. These light-controlled soft switches exhibit strong photothermal responsiveness (~12 s), high programmability, and excellent cyclic stability (>500 times). This research provides a new perspective on light-controlled logic soft switches and their applications in logic circuits. Full article

(This article belongs to the Section Optical Sensors)

12 pages, 6450 KiB

Open AccessArticle

An Electrochemical Nickel–Cobalt (Ni–Co)/Graphene Oxide-Polyvinyl Alcohol (GO-PVA) Sensor for Glucose Detection

by Shu-Hui Yeh, Yaw-Jen Chang and Chun-Yi Hsieh

Sensors 2025, 25(7), 2050; https://doi.org/10.3390/s25072050 - 25 Mar 2025

Abstract

This paper presents a non-enzymatic sensor for glucose detection in an environment where glucose and insulin coexist. The sensor is based on a three-electrode chip fabricated by etching the copper foil of a printed circuit board. The working electrode is coated with a [...] Read more.

This paper presents a non-enzymatic sensor for glucose detection in an environment where glucose and insulin coexist. The sensor is based on a three-electrode chip fabricated by etching the copper foil of a printed circuit board. The working electrode is coated with a graphene oxide-polyvinyl alcohol composite film, followed by the electroplating of a nickel–cobalt layer and an additional surface treatment using O₂ plasma. The experimental results indicate that within a glucose concentration of 2 mM to 10 mM and an insulin concentration of 0.1 mM to 1 mM, the measured current exhibits a linear relationship with the concentration of glucose or insulin, regardless of whether cyclic voltammetry or linear sweep voltammetry is used. However, the detection limit for insulin is 0.01 mM, ensuring that glucose detection remains unaffected by insulin interference. In this sensor, nickel–cobalt serves as a catalyst for glucose and insulin detection, while the graphene oxide-polyvinyl alcohol composite enhances sensing performance. Full article

(This article belongs to the Special Issue Electrochemical Sensors and Biosensors: Materials, Methods and Applications)

► Show Figures

Figure 1

18 pages, 6700 KiB

Open AccessArticle

NightHawk: A Low-Cost, Nighttime Light Wildfire Observation Platform and Its Radiometric Calibration

by Chase A. Fuller, Steve Tammes, Philip Kaaret, Jun Wang, Carlton H. Richey, Marc Linderman, Emmett J. Ientilucci, Thomas Schnell, William Julstrom, Jarret McElrath, Will Meiners, Jack Kelley and Francis Mawanda, Jr.

Sensors 2025, 25(7), 2049; https://doi.org/10.3390/s25072049 - 25 Mar 2025

Abstract

We present a low-cost prototype of a visible and near-infrared (VIS-NIR) remote sensing platform, optimized to detect and characterize natural flaming fire fronts from airborne nighttime light (NTL) observations, and its radiometric calibration. It uses commercially available CMOS sensor cameras and filters with [...] Read more.

We present a low-cost prototype of a visible and near-infrared (VIS-NIR) remote sensing platform, optimized to detect and characterize natural flaming fire fronts from airborne nighttime light (NTL) observations, and its radiometric calibration. It uses commercially available CMOS sensor cameras and filters with roughly 100 nm bandwidths to effectively discriminate burning biomass from other sources of NTL, a critical ability for wildfire monitoring near populated areas. Our filter choice takes advantage of the strong potassium line emission near 770 nm present in natural flaming. The calibrated cameras operate at 20 ms of exposure time and boast radiance measurements with a sensitivity floor, depending on the filter, in the range 3–5

\times 10^{- 6}

W m⁻² sr⁻¹ nm⁻¹ with uncertainties lower than 5% and dynamic ranges near 3000–4000. An additional exposure time with a tenth of the duration is calibrated and extends the dynamic range by a factor of 10. We show images of a spatially resolved fire front from an airborne observation of flaming biomass within this radiance range. Full article

(This article belongs to the Section Remote Sensors)

► Show Figures

Figure 1

27 pages, 5015 KiB

Open AccessArticle

Minima-YOLO: A Lightweight Identification Method for Lithium Mineral Components Under a Microscope Based on YOLOv8

by Zeyang Qiu, Xueyu Huang and Xiangyu Xu

Sensors 2025, 25(7), 2048; https://doi.org/10.3390/s25072048 - 25 Mar 2025

Abstract

Mineral identification technology is a critical technology in the construction of smart mines. To enable effective deployment and implementation of rapid mineral sorting for valuable ores on edge computing devices, we propose a lightweight identification method for lithium minerals under visible light microscopy [...] Read more.

Mineral identification technology is a critical technology in the construction of smart mines. To enable effective deployment and implementation of rapid mineral sorting for valuable ores on edge computing devices, we propose a lightweight identification method for lithium minerals under visible light microscopy based on YOLOv8, named Minima-YOLO. First, by scaling down and limiting the number of channels in the YOLOv8 backbone, we introduced a smaller network, YOLOv8-tiny. Next, we redesigned a new lightweight feature extraction module, Faster-EMA, using PConv and the EMA attention mechanism, replacing the original C2f module. Third, we incorporated GhostConv, a cost-effective downsampling method, as a replacement for standard convolutions. Finally, to mitigate the impact of deeper backbone network layers, we introduced the Slim-Neck structure in the Neck, further reducing the model size. Ultimately, Minima-YOLO achieves a 99.4% mAP50 on our self-constructed lithium mineral image dataset, with FLOPs reduced to 2.3 G, parameters to 0.72 M, and a model size of just 1.63 MB, while maintaining an FPS of 103. A series of comparative experiments confirmed its superior performance over other advanced object detection algorithms. This algorithm, with its highly efficient lightweight design and rapid inference speed, provides an intelligent, efficient, and eco-friendly computer vision method for the rapid sorting of lithium mineral components. Full article

(This article belongs to the Section Sensing and Imaging)

► Show Figures

Figure 1

17 pages, 19409 KiB

Open AccessArticle

Wavelet-Based Topological Loss for Low-Light Image Denoising

by Alexandra Malyugina, Nantheera Anantrasirichai and David Bull

Sensors 2025, 25(7), 2047; https://doi.org/10.3390/s25072047 - 25 Mar 2025

Abstract

Despite significant advances in image denoising, most algorithms rely on supervised learning, with their performance largely dependent on the quality and diversity of training data. It is widely assumed that digital image distortions are caused by spatially invariant Additive White Gaussian Noise (AWGN). [...] Read more.

Despite significant advances in image denoising, most algorithms rely on supervised learning, with their performance largely dependent on the quality and diversity of training data. It is widely assumed that digital image distortions are caused by spatially invariant Additive White Gaussian Noise (AWGN). However, the analysis of real-world data suggests that this assumption is invalid. Therefore, this paper tackles image corruption by real noise, providing a framework to capture and utilise the underlying structural information of an image along with the spatial information conventionally used for deep learning tasks. We propose a novel denoising loss function that incorporates topological invariants and is informed by textural information extracted from the image wavelet domain. The effectiveness of this proposed method was evaluated by training state-of-the-art denoising models on the BVI-Lowlight dataset, which features a wide range of real noise distortions. Adding a topological term to common loss functions leads to a significant increase in the LPIPS (Learned Perceptual Image Patch Similarity) metric, with the improvement reaching up to 25%. The results indicate that the proposed loss function enables neural networks to learn noise characteristics better. We demonstrate that they can consequently extract the topological features of noise-free images, resulting in enhanced contrast and preserved textural information. Full article

(This article belongs to the Special Issue Machine Learning in Image/Video Processing and Sensing)

► Show Figures

Figure 1

16 pages, 1575 KiB

Open AccessArticle

Development of a Piezoceramic Harvester for Sea Waves Energy Recovery in Environmental Monitoring Buoys

by Roberto Montanini, Antonio Cannuli, Fabrizio Freni, Antonino Quattrocchi and Andrea Venuto

Sensors 2025, 25(7), 2046; https://doi.org/10.3390/s25072046 - 25 Mar 2025

Abstract

In the last decades, marine environment monitoring has gained significant attention as it plays a fundamental role in ecosystem health and anthropogenic impact evaluation. This study presents the development of a sea wave energy recovery device based on piezoceramic harvesting, designed to contribute [...] Read more.

In the last decades, marine environment monitoring has gained significant attention as it plays a fundamental role in ecosystem health and anthropogenic impact evaluation. This study presents the development of a sea wave energy recovery device based on piezoceramic harvesting, designed to contribute to the energy self-sufficiency of an environmental monitoring buoy. The system consists of a flexible S-shaped arm anchored to the buoy structure; the buoyancy system at the free end converts wave-induced motion into mechanical stress, deforming the opposite side of the arm, where piezoceramic patches are installed to generate electrical power. An extensive experimental campaign was conducted to perform the electromechanical characterization of the device and to analyze the manufacturing quality of the arm, produced by stereolithographic additive manufacturing. The results demonstrate the ability to harvest kinetic energy across a range of wave frequencies and amplitudes. Under the best conditions, a maximum transfer electric power of 220.2 ± 3.7 µW was reached. Full article

(This article belongs to the Special Issue Piezoelectric Energy Harvesting System)

17 pages, 3512 KiB

Open AccessArticle

Measurement of TBM Disc Cutter Wear Using Eddy-Current Sensor in Different TBM Chamber Conditions: Insights from Laboratory Tests

by Minsung Park, Minseok Ju, Jungjoo Kim and Hoyoung Jeong

Sensors 2025, 25(7), 2045; https://doi.org/10.3390/s25072045 - 25 Mar 2025

Abstract

The TBM disc cutter, which is the main cutting tool of tunnel boring machines (TBMs), is replaced when it is excessively worn during the boring process. Disc cutters are usually monitored by workers at cutterhead chambers, and they check the status and wear [...] Read more.

The TBM disc cutter, which is the main cutting tool of tunnel boring machines (TBMs), is replaced when it is excessively worn during the boring process. Disc cutters are usually monitored by workers at cutterhead chambers, and they check the status and wear of disc cutters. Manual measurement occasionally results in inaccurate measurement results. In order to overcome these limitations, real-time disc cutter monitoring techniques have been developed with different types of sensors. This study evaluates the distance measurement performance of an eddy-current sensor for measuring disc cutter wear via a series of laboratory experiments. This study focused on identifying the effects of various measurement environments on the sensor’s accuracy. The study considered conditions that the eddy-current sensor may encounter in shield TBM chambers, including air, water, slurry, and excavated muck. Experiments were conducted using both a small-scale disc cutter and a 17-inch full-scale disc cutter. The results indicate that the eddy-current sensor can accurately measure the distance to the disc cutter within a specific range and that its performance remains unaffected by different measurement environments. Full article

(This article belongs to the Special Issue Advanced Sensors for Monitoring and Detection in Geotechnical Engineering)

21 pages, 936 KiB

Open AccessArticle

Integrating Color and Contour Analysis with Deep Learning for Robust Fire and Smoke Detection

by Abror Shavkatovich Buriboev, Akmal Abduvaitov and Heung Seok Jeon

Sensors 2025, 25(7), 2044; https://doi.org/10.3390/s25072044 - 25 Mar 2025

Abstract

Detecting fire and smoke is essential for maintaining safety in urban, industrial, and outdoor settings. This study suggests a unique concatenated convolutional neural network (CNN) model that combines deep learning with hybrid preprocessing methods, such as contour-based algorithms and color characteristics analysis, to [...] Read more.

Detecting fire and smoke is essential for maintaining safety in urban, industrial, and outdoor settings. This study suggests a unique concatenated convolutional neural network (CNN) model that combines deep learning with hybrid preprocessing methods, such as contour-based algorithms and color characteristics analysis, to provide reliable and accurate fire and smoke detection. A benchmark dataset with a variety of situations, including dynamic surroundings and changing illumination, the D-Fire dataset was used to assess the technique. Experiments show that the suggested model outperforms both conventional techniques and the most advanced YOLO-based methods, achieving accuracy (0.989) and recall (0.983). In order to reduce false positives and false negatives, the hybrid architecture uses preprocessing to enhance Regions of Interest (ROIs). Additionally, pooling and fully linked layers provide computational efficiency and generalization. In contrast to current approaches, which frequently concentrate only on fire detection, the model’s dual smoke and fire detection capabilities increase its adaptability. Although preprocessing adds a little computing expense, the methodology’s excellent accuracy and resilience make it a dependable option for safety-critical real-world applications. This study sets a new standard for smoke and fire detection and provides a route forward for future developments in this crucial area. Full article

(This article belongs to the Special Issue Intelligent Sensing and Artificial Intelligence for Image Processing)

29 pages, 4530 KiB

Open AccessSystematic Review

Advances in Deep Learning for Semantic Segmentation of Low-Contrast Images: A Systematic Review of Methods, Challenges, and Future Directions

by Claudio Urrea and Maximiliano Vélez

Sensors 2025, 25(7), 2043; https://doi.org/10.3390/s25072043 - 25 Mar 2025

Abstract

The semantic segmentation (SS) of low-contrast images (LCIs) remains a significant challenge in computer vision, particularly for sensor-driven applications like medical imaging, autonomous navigation, and industrial defect detection, where accurate object delineation is critical. This systematic review develops a comprehensive evaluation of state-of-the-art [...] Read more.

The semantic segmentation (SS) of low-contrast images (LCIs) remains a significant challenge in computer vision, particularly for sensor-driven applications like medical imaging, autonomous navigation, and industrial defect detection, where accurate object delineation is critical. This systematic review develops a comprehensive evaluation of state-of-the-art deep learning (DL) techniques to improve segmentation accuracy in LCI scenarios by addressing key challenges such as diffuse boundaries and regions with similar pixel intensities. It tackles primary challenges, such as diffuse boundaries and regions with similar pixel intensities, which limit conventional methods. Key advancements include attention mechanisms, multi-scale feature extraction, and hybrid architectures combining Convolutional Neural Networks (CNNs) with Vision Transformers (ViTs), which expand the Effective Receptive Field (ERF), improve feature representation, and optimize information flow. We compare the performance of 25 models, evaluating accuracy (e.g., mean Intersection over Union (mIoU), Dice Similarity Coefficient (DSC)), computational efficiency, and robustness across benchmark datasets relevant to automation and robotics. This review identifies limitations, including the scarcity of diverse, annotated LCI datasets and the high computational demands of transformer-based models. Future opportunities emphasize lightweight architectures, advanced data augmentation, integration with multimodal sensor data (e.g., LiDAR, thermal imaging), and ethically transparent AI to build trust in automation systems. This work contributes a practical guide for enhancing LCI segmentation, improving mean accuracy metrics like mIoU by up to 15% in sensor-based applications, as evidenced by benchmark comparisons. It serves as a concise, comprehensive guide for researchers and practitioners advancing DL-based LCI segmentation in real-world sensor applications. Full article

(This article belongs to the Section Sensing and Imaging)

► Show Figures

Figure 1

31 pages, 2350 KiB

Open AccessReview

Emerging Applications of Augmented and Mixed Reality Technologies in Motor Rehabilitation: A Scoping Review

by Arman Farsi, Giacinto Luigi Cerone, Deborah Falla and Marco Gazzoni

Sensors 2025, 25(7), 2042; https://doi.org/10.3390/s25072042 (registering DOI) - 25 Mar 2025

Abstract

Background: Augmented Reality (AR) and Mixed Reality (MR) are emerging technologies with notable potential for motor rehabilitation. Given the novelty and breadth of this field, this scoping review aims to identify how and to what extent AR and MR technologies are used in [...] Read more.

Background: Augmented Reality (AR) and Mixed Reality (MR) are emerging technologies with notable potential for motor rehabilitation. Given the novelty and breadth of this field, this scoping review aims to identify how and to what extent AR and MR technologies are used in motor rehabilitation. Methods: We conducted a search in Scopus and PubMed (2010–2024), following PRISMA-ScR guidelines. In the analysis, we focused on four key aspects: (I) the AR/MR display technologies, (II) the sensors used to collect data to generate the augmented information, (III) the pathologies addressed, and (IV) the assessment of usability and acceptability. Results: Among 105 selected studies, 58% developed new prototypes, while 42% tested existing systems. Head-mounted displays were the most common device (56.2%), followed by monitors (34.3%) and video projectors (14.3%). The most commonly used sensors were RGB-D cameras (31.4%), sensors for localization and mapping (33.3%), normal cameras (17.1%), and electromyography sensors (14.3%). Regarding the target pathology, 34.2% of studies did not focus on a specific pathology, 26.7% were on stroke, 10.5% on limb loss, and 9.5% on Parkinson’s disease. Over half (51.4%) of the studies investigated usability and acceptance. Conclusions: AR/MR technologies hold promise for motor rehabilitation, but limited comparative studies and long-term investigations currently hinder a clear understanding of their benefits. Full article

(This article belongs to the Special Issue Extended Reality in Medicine and Healthcare: Methods, Technologies, Applications and Future Trends)

► Show Figures

Figure 1

33 pages, 6859 KiB

Open AccessArticle

An Improved Autonomous Emergency Braking Algorithm for AGVs: Enhancing Operational Smoothness Through Multi-Stage Deceleration

by Wenbo Li and Junting Qiu

Sensors 2025, 25(7), 2041; https://doi.org/10.3390/s25072041 - 25 Mar 2025

Abstract

The automated guided vehicle (AGV) is widely used in industrial environments for goods transportation. However, issues such as mechanical wear, reduced battery life, navigation error accumulation, and decreased operational efficiency caused by frequent starts and stops need to be addressed. This paper proposes [...] Read more.

The automated guided vehicle (AGV) is widely used in industrial environments for goods transportation. However, issues such as mechanical wear, reduced battery life, navigation error accumulation, and decreased operational efficiency caused by frequent starts and stops need to be addressed. This paper proposes an improved Autonomous Emergency Braking (AEB) algorithm to tackle these problems. The algorithm employs a stepwise deceleration strategy, effectively reducing the frequency of sudden stops and enhancing the system’s operational smoothness. The AEB algorithm not only considers straight-line driving scenarios but also optimizes deceleration strategies for turning scenarios, adjusting the deceleration detection range according to the turning trajectory. Additionally, a velocity smoothing algorithm is designed to ensure that speed changes during deceleration are gradual, avoiding abrupt speed variations that could impact the system. The feasibility of the AEB algorithm is validated through testing on actual equipment, and its performance is compared to that of a conventional emergency stop strategy. Experimental results show that the AEB algorithm significantly reduces the number of sudden stops, improves the AGV’s operational smoothness and safety, and demonstrates excellent adaptability and robustness across different operational conditions. Full article

(This article belongs to the Section Industrial Sensors)

► Show Figures

Figure 1

20 pages, 1958 KiB

Open AccessReview

Continuous Wave-Diffuse Optical Tomography (CW-DOT) in Human Brain Mapping: A Review

by Shuo Guan, Yuhang Li, Yuanyuan Gao, Yuxi Luo, Hubin Zhao, Dalin Yang and Rihui Li

Sensors 2025, 25(7), 2040; https://doi.org/10.3390/s25072040 (registering DOI) - 25 Mar 2025

Abstract

Continuous wave-diffuse optical tomography (CW-DOT) has emerged as a promising non-invasive neuroimaging technique for assessing brain function. Its ability to provide brain mapping with high spatial resolution over traditional functional near-infrared spectroscopy (fNIRS) has garnered significant interest in clinical and cognitive neuroscience. In [...] Read more.

Continuous wave-diffuse optical tomography (CW-DOT) has emerged as a promising non-invasive neuroimaging technique for assessing brain function. Its ability to provide brain mapping with high spatial resolution over traditional functional near-infrared spectroscopy (fNIRS) has garnered significant interest in clinical and cognitive neuroscience. In this review, we critically summarized the hardware, reconstruction algorithms, and applications of CW-DOT for human brain mapping, providing an up-to-date overview and guidelines for future studies to conduct CW-DOT studies. ScienceDirect, PubMed, Web of Science, and IEEE Xplore databases were searched from their inception up to 1 July 2024. A total of 83 articles were included in the final systematic review. The review focused on existing hardware systems, reconstruction algorithms for CW-DOT, and the applications of CW-DOT in both clinical settings and cognitive neuroscience. Finally, we highlighted current challenges and potential directions of CW-DOT in future research, including the absence of standardized protocols and a pressing need for enhanced quantitative precision. This review underscores the sophisticated capabilities of CW-DOT systems, particularly in the realm of human brain imaging. Extensive clinical and neuroscience research has attested to the technique’s anatomical precision and reliability, establishing it as a potent instrument in research and clinical practice. Full article

(This article belongs to the Section Biomedical Sensors)

► Show Figures

Figure 1

29 pages, 1768 KiB

Open AccessArticle

Deep Reinforcement Learning-Enabled Computation Offloading: A Novel Framework to Energy Optimization and Security-Aware in Vehicular Edge-Cloud Computing Networks

by Waleed Almuseelem

Sensors 2025, 25(7), 2039; https://doi.org/10.3390/s25072039 (registering DOI) - 25 Mar 2025

Abstract

The Vehicular Edge-Cloud Computing (VECC) paradigm has gained traction as a promising solution to mitigate the computational constraints through offloading resource-intensive tasks to distributed edge and cloud networks. However, conventional computation offloading mechanisms frequently induce network congestion and service delays, stemming from uneven [...] Read more.

The Vehicular Edge-Cloud Computing (VECC) paradigm has gained traction as a promising solution to mitigate the computational constraints through offloading resource-intensive tasks to distributed edge and cloud networks. However, conventional computation offloading mechanisms frequently induce network congestion and service delays, stemming from uneven workload distribution across spatial Roadside Units (RSUs). Moreover, ensuring data security and optimizing energy usage within this framework remain significant challenges. To this end, this study introduces a deep reinforcement learning-enabled computation offloading framework for multi-tier VECC networks. First, a dynamic load-balancing algorithm is developed to optimize the balance among RSUs, incorporating real-time analysis of heterogeneous network parameters, including RSU computational load, channel capacity, and proximity-based latency. Additionally, to alleviate congestion in static RSU deployments, the framework proposes deploying UAVs in high-density zones, dynamically augmenting both storage and processing resources. Moreover, an Advanced Encryption Standard (AES)-based mechanism, secured with dynamic one-time encryption key generation, is implemented to fortify data confidentiality during transmissions. Further, a context-aware edge caching strategy is implemented to preemptively store processed tasks, reducing redundant computations and associated energy overheads. Subsequently, a mixed-integer optimization model is formulated that simultaneously minimizes energy consumption and guarantees latency constraint. Given the combinatorial complexity of large-scale vehicular networks, an equivalent reinforcement learning form is given. Then a deep learning-based algorithm is designed to learn close-optimal offloading solutions under dynamic conditions. Empirical evaluations demonstrate that the proposed framework significantly outperforms existing benchmark techniques in terms of energy savings. These results underscore the framework’s efficacy in advancing sustainable, secure, and scalable intelligent transportation systems. Full article

(This article belongs to the Special Issue Vehicle-to-Everything (V2X) Communication Networks 2024–2025)

► Show Figures

Figure 1

18 pages, 2108 KiB

Open AccessArticle

A Lightweight Approach to Comprehensive Fabric Anomaly Detection Modeling

by Shuqin Cui, Weihong Liu and Min Li

Sensors 2025, 25(7), 2038; https://doi.org/10.3390/s25072038 (registering DOI) - 25 Mar 2025

Abstract

In order to solve the problem of high computational resource consumption in fabric anomaly detection, we propose a lightweight network, GH-YOLOx, which integrates ghost convolutions and hierarchical GHNetV2 backbone together to capture both local and global anomaly features. At the same time, other [...] Read more.

In order to solve the problem of high computational resource consumption in fabric anomaly detection, we propose a lightweight network, GH-YOLOx, which integrates ghost convolutions and hierarchical GHNetV2 backbone together to capture both local and global anomaly features. At the same time, other innovative components, such as GhostConv, dynamic convolutions, feature fusion modules, and a shared group convolution head, are applied to effectively handle multi-scale issues. Lamp pruning accelerates inference, while channel-wise knowledge distillation enhances the pruned model’s accuracy. Experiments on fabric datasets demonstrate that GH-YOLOx can effectively reduce the number of parameters while achieving a higher detection rate than other lightweight models. Overall, our solution offers a practical approach to real-time fabric anomaly detection on mobile and embedded devices. Full article

(This article belongs to the Section Intelligent Sensors)

► Show Figures

Figure 1

17 pages, 5771 KiB

Open AccessArticle

RelVid: Relational Learning with Vision-Language Models for Weakly Video Anomaly Detection

by Jingxin Wang, Guohan Li, Jiaqi Liu, Zhengyi Xu, Xinrong Chen and Jianming Wei

Sensors 2025, 25(7), 2037; https://doi.org/10.3390/s25072037 (registering DOI) - 25 Mar 2025

Abstract

Weakly supervised video anomaly detection aims to identify abnormal events in video sequences without requiring frame-level supervision, which is a challenging task in computer vision. Traditional methods typically rely on low-level visual features with weak supervision from a single backbone branch, which often [...] Read more.

Weakly supervised video anomaly detection aims to identify abnormal events in video sequences without requiring frame-level supervision, which is a challenging task in computer vision. Traditional methods typically rely on low-level visual features with weak supervision from a single backbone branch, which often struggles to capture the distinctive characteristics of different categories. This limitation reduces their adaptability to real-world scenarios. In real-world situations, the boundary between normal and abnormal events is often unclear and context-dependent. For example, running on a track may be considered normal, but running on a busy road could be deemed abnormal. To address these challenges, RelVid is introduced as a novel framework that improves anomaly detection by expanding the relative feature gap between classes extracted from a single backbone branch. The key innovation of RelVid lies in the integration of auxiliary tasks, which guide the model to learn more discriminative features, significantly boosting the model’s performance. These auxiliary tasks—including text-based anomaly detection and feature reconstruction learning—act as additional supervision, helping the model capture subtle differences and anomalies that are often difficult to detect in weakly supervised settings. In addition, RelVid incorporates two other components, which include class activation feature learning for improved feature discrimination and a temporal attention module for capturing sequential dependencies. This approach enhances the model’s robustness and accuracy, enabling it to better handle complex and ambiguous scenarios. Evaluations on two widely used benchmark datasets, UCF-Crime and XD-Violence, demonstrate the effectiveness of RelVid. Compared to state-of-the-art methods, RelVid achieves superior performance in both detection accuracy and robustness. Full article

(This article belongs to the Section Intelligent Sensors)

► Show Figures

Figure 1

15 pages, 3691 KiB

Open AccessArticle

Adaptive Steered Frequency–Wavenumber Analysis for High-Frequency Source Localization in Shallow Water

by Y. H. Choi, Gihoon Byun, Donghyeon Kim and J. S. Kim

Sensors 2025, 25(7), 2036; https://doi.org/10.3390/s25072036 (registering DOI) - 25 Mar 2025

Abstract

In shallow-water environments, source localization often suffers from reduced performance in conventional array signal processing techniques for frequency bands above 1 kHz due to environmental mismatch. A recently proposed technique, called the steered frequency–wavenumber (SFK) analysis method, overcomes this limitation. By incorporating beam-steering [...] Read more.

In shallow-water environments, source localization often suffers from reduced performance in conventional array signal processing techniques for frequency bands above 1 kHz due to environmental mismatch. A recently proposed technique, called the steered frequency–wavenumber (SFK) analysis method, overcomes this limitation. By incorporating beam-steering techniques into frequency–wavenumber analysis, this method enables target localization even in sparse conditions where high-frequency signals are received. This study extends the SFK method by applying various adaptive signal processing techniques, with a particular focus on the minimum-variance distortionless response and white noise gain constraint methods. Using snapping shrimp sounds from the SAVEX15 experiment, we analyzed localization performance and compared it with the Bartlett SFK approach. The snapping shrimp signals have frequency components ranging from 5 to 24 kHz and exhibit impulsive characteristics with a duration of 0.2 ms. Signals recorded by a sparse vertical array of 16 sensors, with a 60-m aperture in 100-m shallow water, enabled the localization of a source at a range of 38 m and a depth of 99.8 m. Full article

(This article belongs to the Section Environmental Sensing)

► 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