Sign in to use this feature.

Years

Between: -

Subjects

remove_circle_outline
remove_circle_outline
remove_circle_outline
remove_circle_outline
remove_circle_outline
remove_circle_outline
remove_circle_outline

Journals

Article Types

Countries / Regions

Search Results (64)

Search Parameters:
Keywords = receptive field correspondence

Order results
Result details
Results per page
Select all
Export citation of selected articles as:
20 pages, 2795 KB  
Article
PES-PointPillars: LiDAR-Based 3D Object Detection for Autonomous Driving with Directional Convolution, Adaptive Feature Fusion, and Decoupled Regression
by Yanbo Song and Meichen Liu
Electronics 2026, 15(17), 3767; https://doi.org/10.3390/electronics15173767 - 22 Aug 2026
Viewed by 93
Abstract
LiDAR-based 3D object detection for autonomous driving must balance localization accuracy with real-time inference, while sparse point measurements make small-scale objects such as pedestrians and cyclists particularly challenging to represent at long range. This paper presents PES-PointPillars, an enhanced PointPillars detector with three [...] Read more.
LiDAR-based 3D object detection for autonomous driving must balance localization accuracy with real-time inference, while sparse point measurements make small-scale objects such as pedestrians and cyclists particularly challenging to represent at long range. This paper presents PES-PointPillars, an enhanced PointPillars detector with three coordinated design changes. First, pinwheel-shaped convolution (PConv) replaces selected backbone convolutions to expand horizontal and vertical receptive fields for sparse structural patterns. Second, an Improved Inter-Layer Feature Correlation (I-EFC) module uses soft gating and adaptive thresholding to fuse multi-level features through continuous, input-dependent weights. Third, a Smooth L1-NWD (SNWD) loss applies normalized Wasserstein distance to planar position and scale while retaining Smooth L1 regression for vertical position, height, and orientation. Using the parameter settings and configuration of the original PointPillars implementation, the locally executed PES-PointPillars experiment achieves Moderate 3D average precision values of 77.1% for cars, 46.7% for pedestrians, and 62.9% for cyclists at 68.3 FPS on the KITTI validation split. Relative to the source-reported PointPillars reference, the corresponding numerical differences are 2.1, 3.2, and 3.8 percentage points. The reported component-wise and staged ablations show category-dependent gains, with the complete model providing the strongest aggregate performance among the evaluated configurations. Full article
(This article belongs to the Special Issue Feature Papers in Electrical and Autonomous Vehicles, Volume 2)
Show Figures

Figure 1

21 pages, 2796 KB  
Article
FD-GhostFaceNet: Frequency-Decoupled Ghost Modules for Lightweight Face Recognition
by Abdalbasit Qadir, Bryar A. Hassan and Hozan Khalid
Computers 2026, 15(8), 514; https://doi.org/10.3390/computers15080514 - 7 Aug 2026
Viewed by 294
Abstract
Deploying accurate face recognition models on resource-constrained devices remains a significant challenge. Ghost modules mitigate feature-map redundancy by synthesizing half of each feature map through inexpensive linear transformations and form the basis of the lightweight GhostFaceNet and GhostFaceNet++ families. Nevertheless, the inexpensive generator [...] Read more.
Deploying accurate face recognition models on resource-constrained devices remains a significant challenge. Ghost modules mitigate feature-map redundancy by synthesizing half of each feature map through inexpensive linear transformations and form the basis of the lightweight GhostFaceNet and GhostFaceNet++ families. Nevertheless, the inexpensive generator is a strictly local 3 × 3 depthwise convolution: the synthesized features remain near-duplicates of the intrinsic ones, and deeper stages lack the global receptive field required for cross-pose and cross-age matching. This paper introduces FD-GhostFaceNet, built upon the proposed Frequency-Decoupled Ghost (FD-Ghost) module, which replaces the local generator with a complementary-band global counterpart. A learnable, input-conditioned partition of the two-dimensional Discrete Cosine Transform spectrum decomposes the intrinsic features into a global low-frequency band that encodes pose- and age-stable identity structure and a locally refined high-frequency detail band. Nyquist-consistent anti-aliased downsampling, which band-limits each feature map before subsampling so that the decoupled high-frequency band survives decimation, completes the frequency-decoupled design. Applied to both published trunks, FD-GhostFaceNet-V1-2 and FD-GhostFaceNet-V2-2 require 4.60 M and 7.39 M parameters, 83.1 and 97.1 MFLOPs, and 9.19 and 14.77 MB of storage, respectively, and thus fall within the sub-100 MFLOPs category of lightweight face recognition models. Across six standard benchmarks, both variants surpass their corresponding GhostFaceNet baselines in 23 of 24 trunk–benchmark comparisons and are competitive with or superior to the GhostFaceNet++ variants, with the largest gains on the pose-sensitive evaluations. When trained with ArcFace on UMDFaces, FD-GhostFaceNet raises the best reported CP-LFW accuracy from 84.65% to 86.12% and CFP-FP from 87.60% to 89.56%; when trained on CASIA-WebFace, it improves the best reported CFP-FP from 90.10% to 90.94%. These results confirm that frequency decoupling strikes a favorable balance between compactness and accuracy, making both models strong candidates for deployment on resource-constrained devices. Full article
(This article belongs to the Section AI-Driven Innovations)
Show Figures

Figure 1

18 pages, 1924 KB  
Article
Context-Gated Graph Modelling for Traffic Flow Forecasting
by Yuzhuo Zhang, Jialin Liang, Ziqiong Yuan, Zanzan Dai and Yaozheng Kang
Modelling 2026, 7(4), 157; https://doi.org/10.3390/modelling7040157 - 5 Aug 2026
Viewed by 162
Abstract
Traffic states evolve on irregular sensor graphs and vary with calendar context, yet the original ASTGCN does not explicitly model how the contribution of different graph receptive fields changes across traffic periods. This paper proposes CD-MRFG, a context-gated extension of ASTGCN that encodes [...] Read more.
Traffic states evolve on irregular sensor graphs and vary with calendar context, yet the original ASTGCN does not explicitly model how the contribution of different graph receptive fields changes across traffic periods. This paper proposes CD-MRFG, a context-gated extension of ASTGCN that encodes hour-of-day, day-of-week and weekend information and uses the resulting representation to weight Chebyshev graph-convolution orders in each spatio-temporal block. Under a common 12-step forecasting protocol, CD-MRFG reduced the overall MAE and RMSE of the reproduced ASTGCN baseline from 18.66 and 31.05 to 16.98 and 28.59 on PEMS03, from 22.79 and 35.02 to 20.82 and 32.77 on PEMS04, and from 18.88 and 28.83 to 17.24 and 26.84 on PEMS08. Three-seed experiments confirmed lower mean MAEs on PEMS04 (p = 0.028) and PEMS08 (p = 0.042), although the corresponding RMSE differences did not reach the 0.05 significance threshold. Ablation, gate-weight, sensitivity, complexity and convergence analyses showed that temporal context was the main source of the improvement and that the gate provided a model-internal view of order selection with moderate overhead. CD-MRFG remains less accurate than several stronger recent baselines, so its value is a bounded and interpretable extension of ASTGCN rather than a universal state-of-the-art replacement. Full article
(This article belongs to the Section Modelling in Artificial Intelligence)
Show Figures

Figure 1

37 pages, 29029 KB  
Article
High-Precision Flood Extraction from High-Resolution Remote Sensing Images by Integrating FCN-RAM and Tolerance Rough Set
by Ximin Yuan, Haotian Xu, Xiujie Wang and Fuchang Tian
Remote Sens. 2026, 18(14), 2373; https://doi.org/10.3390/rs18142373 - 16 Jul 2026
Viewed by 399
Abstract
High-precision flood identification from high-resolution remote sensing images using deep learning network models is challenging. Severe cloud interference, limited receptive fields, insufficient boundary refinement and spatial detail preservation, and difficulty in accurately distinguishing water bodies from ground object shadows constrain the extraction method. [...] Read more.
High-precision flood identification from high-resolution remote sensing images using deep learning network models is challenging. Severe cloud interference, limited receptive fields, insufficient boundary refinement and spatial detail preservation, and difficulty in accurately distinguishing water bodies from ground object shadows constrain the extraction method. Therefore, this study proposes an automatic flood information extraction method that integrates an improved Fully Convolutional Network classification and recognition model (FCN-RAM) with a rough tolerance set. First, a tolerance rough set algorithm was employed for sample data preprocessing. Subsequently, a Residual Attention Module (RAM) was introduced to optimize the U-Net architecture, dynamically adjusting the response intensity of deep features in both the channel and spatial dimensions to construct a deep learning-based FCN-RAM. Finally, comparative analyses were conducted on three high-resolution remote sensing datasets with different resolutions: Global surface water detection in Large-size very-High-resolution satellite imagery (GLH-Water), Gaofen Image Dataset (GID), and Earth Surface Water Dataset (ESWD). The results demonstrated that FCN-RAM consistently and substantially outperformed the baseline U-Net across all three datasets, achieving F1-score improvements of 10.64% (GLH-Water), 9.71% (GID), and 10.64% (ESWD), with corresponding overall accuracy gains of 9.97%, 11.15%, and 10.22%, respectively. Notably, the Intersection-over-Union (IoU) scores were elevated by 17.59% (GLH-Water), 15.66% (GID), and 13.63% (ESWD). The method also surpassed state-of-the-art models including ResNet and Water-SCNet, attaining peak overall accuracies of 98.61% (GLH-Water) and 97.37% (GID). Notably, while the proposed framework exhibits remarkable generalization across the evaluated multi-resolution benchmarks, its current validation is primarily confined to static water body delineation tasks. The model’s transferability to highly heterogeneous geographical regions with scarce training samples, as well as its extendability toward dynamic time-series flood evolution modeling, warrants further systematic investigation. The proposed method significantly improves the accuracy of waterbody information extraction, meets the requirements for high-precision information extraction from high-resolution imagery, and provides technical support for intelligent flood information extraction using high-resolution remote sensing. Full article
Show Figures

Figure 1

30 pages, 13264 KB  
Article
Designof a Quasi-Real-Time OCDM Underwater Communication System Based on a Software-Defined Architecture
by Jiali Chen, Zhenquan Hu, Wen Chen, Peibin Zhu and Xiaomei Xu
J. Mar. Sci. Eng. 2026, 14(14), 1285; https://doi.org/10.3390/jmse14141285 - 13 Jul 2026
Viewed by 351
Abstract
Shallow-water underwater acoustic (UWA) communication is strongly affected by long-delay multipath propagation, time-varying Doppler distortion, and the computational burden of coherent demodulation. This paper presents a software-defined, quasi-real-time orthogonal chirp division multiplexing (OCDM) communication prototype implemented on a National Instruments compactRIO heterogeneous platform. [...] Read more.
Shallow-water underwater acoustic (UWA) communication is strongly affected by long-delay multipath propagation, time-varying Doppler distortion, and the computational burden of coherent demodulation. This paper presents a software-defined, quasi-real-time orthogonal chirp division multiplexing (OCDM) communication prototype implemented on a National Instruments compactRIO heterogeneous platform. The prototype maps previously developed multiplex resampling (MR) and Data Pick-Rake (DP-Rake) algorithms to a hardware–software processing chain, in which the FPGA executes sample-level physical-layer operations, including MR, DP-Rake windowing, OCDM demodulation, and equalization, while the ARM real-time controller performs frame-level control, FEC/CRC processing, and state-machine scheduling. The DMA-based RT–FPGA data exchange achieved a measured throughput of 80–100 MB/s. Under the maximum MR configuration, the FPGA physical-layer processing, DMA transfer, and ARM RT-layer processing required approximately 4.5 ms, 0.5 ms, and 5.0 ms per data block, respectively, resulting in a total receiver-side digital processing latency of approximately 10.0 ms. This corresponds to a 32.6 ms timing margin relative to the shortest acoustic data-block duration. The maximum FPGA resource utilization among the reported resource categories was 56.8%. The prototype was evaluated through replay-based channel emulation, Qiandao Lake mobility experiments, and Xiamen Outer Port sea trials. The replay-based results show that DP-Rake reception reduces the multipath-induced error floor under severe delay-spread conditions. In the Qiandao Lake experiments, dynamic MR branch selection achieved zero packet errors within the tested observation window while reducing the average number of active MR branches by 32.7% compared with the fixed-branch configuration. In the Xiamen Outer Port sea trials, an over 6 km shallow-water link with an approximately 12 ms delay spread was evaluated, and no packet errors were observed within the tested observation window when the measured SNR exceeded 7.9 dB. These results indicate that the implemented MR-DP-Rake OCDM prototype improves robustness over the OFDM-based baseline under the tested doubly spread UWA channel conditions. The main contribution of this work is the end-to-end software-defined implementation and field validation of MR-DP-Rake OCDM, rather than the proposal of new communication-theoretic algorithms. Full article
(This article belongs to the Section Ocean Engineering)
Show Figures

Figure 1

19 pages, 7632 KB  
Article
BSEF-Stereo: A Stereo Matching Model Based on Branching Strategy and Error Feedback
by Kaicheng Li, Jinlong Yang and Chin Chi Choi
Sensors 2026, 26(13), 4318; https://doi.org/10.3390/s26134318 - 7 Jul 2026
Viewed by 367
Abstract
Iterative stereo matching remains challenging in weakly textured regions, repetitive patterns, occlusions, and object boundaries, where ambiguous correspondence cues require broad contextual reasoning while accurate reconstruction depends on preserving local structural details. Existing recurrent updaters with a fixed receptive field struggle to balance [...] Read more.
Iterative stereo matching remains challenging in weakly textured regions, repetitive patterns, occlusions, and object boundaries, where ambiguous correspondence cues require broad contextual reasoning while accurate reconstruction depends on preserving local structural details. Existing recurrent updaters with a fixed receptive field struggle to balance these requirements, and their initial disparity estimates may retain local geometric inconsistencies. To address these limitations, we propose BSEF-Stereo, an iterative framework that combines adaptive recurrent updating with explicit error-feedback refinement. A channel–position attention module strengthens discriminative channel and spatial cues, while a branch-strategy gated recurrent unit uses complementary small- and large-kernel branches to preserve boundary details and aggregate context in ambiguous regions. An error-aware refinement module subsequently exploits reprojection error and image guidance to correct the initial disparity map. Experiments on Scene Flow, KITTI 2012, KITTI 2015, and Middlebury demonstrate competitive performance across synthetic, outdoor, and indoor scenes. BSEF-Stereo achieves 0.41 px EPE and 2.27% D1 on Scene Flow and a D1-all error of 1.48% on KITTI 2015. Ablation and sensitivity studies verify the complementary contributions of the three modules and support the selected design settings. Full article
(This article belongs to the Special Issue AI-Based Visual Sensing for Object Detection)
Show Figures

Figure 1

19 pages, 17897 KB  
Article
S2M-Net: Dynamic Hyperspectral Unmixing Network Integrating Spectral Sequence Mamba and Local Spatial–Spectral Awareness
by Yongqing Yang, Mengmeng Xu, Weidong Zhang, Ji Zhang and Yuquan Gan
Remote Sens. 2026, 18(13), 2228; https://doi.org/10.3390/rs18132228 - 6 Jul 2026
Viewed by 477
Abstract
Hyperspectral unmixing aims to extract pure endmembers and their corresponding abundance from mixed pixels. Existing deep learning-based unmixing methods predominantly rely on convolutional neural networks (CNNs) or Transformer architectures. However, CNNs suffer from limited receptive fields and struggle to capture long-range spectral dependencies [...] Read more.
Hyperspectral unmixing aims to extract pure endmembers and their corresponding abundance from mixed pixels. Existing deep learning-based unmixing methods predominantly rely on convolutional neural networks (CNNs) or Transformer architectures. However, CNNs suffer from limited receptive fields and struggle to capture long-range spectral dependencies across the entire spectral sequence. While Transformers possess global modeling capabilities, they are constrained by quadratic computational complexity and lack the ability to adaptively filter redundant noise in consecutive spectral bands. To address these limitations, this paper proposes a dynamic hyperspectral unmixing network integrating a spectral sequence Mamba with local spatial–spectral awareness. Specifically, the network features a novel asymmetric dual-stream collaborative architecture. The first branch, the spectral sequence Mamba, models hyperspectral data as a one-dimensional continuous sequence and employs the selective state space model to perform global scanning with linear complexity. This adaptively filters redundant spectral bands to accurately extract high-purity global spectral semantics. The second branch, dedicated to local spatial–spectral awareness, uses an attention-augmented CNN to capture local continuous spectral variations and spatial textures, providing fine-grained geometric boundary constraints for abundance estimation. Furthermore, a spatially adaptive gated fusion module is designed to dynamically balance global spectral semantics and local spatial–spectral details according to the pixel mixing complexity of varying spatial regions. Extensive experiments on multiple public hyperspectral datasets demonstrate that the proposed method achieves significant improvements in unmixing accuracy over comparative methods. Full article
Show Figures

Figure 1

21 pages, 3551 KB  
Article
Progressive Pixel-Neighborhood Deformable Cross-Attention for Multispectral Object Detection
by Tian Qiu, Jifeng Shen and Xin Zuo
Sensors 2026, 26(12), 3825; https://doi.org/10.3390/s26123825 - 16 Jun 2026
Viewed by 433
Abstract
Effective cross-modal feature alignment and interaction are central challenges in multispectral object detection. Although global cross-attention provides strong long-range modeling ability, its quadratic complexity with respect to feature size limits deployment on resource-constrained platforms. We therefore propose Progressive Pixel-Neighborhood Deformable Cross-Attention for multispectral [...] Read more.
Effective cross-modal feature alignment and interaction are central challenges in multispectral object detection. Although global cross-attention provides strong long-range modeling ability, its quadratic complexity with respect to feature size limits deployment on resource-constrained platforms. We therefore propose Progressive Pixel-Neighborhood Deformable Cross-Attention for multispectral feature fusion, termed PNAFusion. The proposed framework is motivated by two observations: weak misalignment between visible and thermal images is usually concentrated around local neighborhoods, and semantic correspondence across modalities often follows non-linear spatial mappings that fixed receptive fields cannot model well. To address these issues, PNAFusion incorporates local spatial priors into its architectural design to concentrate feature interaction and alignment on the most relevant neighborhoods. Specifically, a Pixel-Neighborhood Cross-Attention (PNCA) module is introduced to avoid redundant global feature matching and suppress background noise. Meanwhile, an Adaptive Deformable Alignment (ADA) module captures non-linear spatial correspondences through learned pixel-wise offsets. These components are further integrated through an iterative feedback mechanism to progressively refine cross-modal feature alignment. Experiments on FLIR, M3FD, and DroneVehicle show that PNAFusion achieves 84.2, 90.5, and 85.5 mAP@0.5, respectively, under the YOLOv5 detector, and further reaches 86.8 mAP@0.5 on FLIR and 90.8 mAP@0.5 on M3FD when transferred to Co-DETR. Efficiency analysis indicates that PNAFusion reduces allocated GPU memory by 33.0% compared with ICAFusion and reduces theoretical FLOPs from 194.8 G to 156.4 G, although the deformable sampling and iterative refinement introduce additional latency. These results demonstrate that PNAFusion provides a practical accuracy–memory trade-off for weakly aligned multispectral object detection. Full article
(This article belongs to the Special Issue Image Processing and Analysis for Object Detection: 3rd Edition)
Show Figures

Figure 1

24 pages, 9966 KB  
Article
A Cross-Layer Feature Fusion Framework with Hierarchical Interaction for Remote Sensing Change Detection
by Xin Meng, Chuanbiao Qiu, Chong Liu and Yanli Xu
Sensors 2026, 26(4), 1176; https://doi.org/10.3390/s26041176 - 11 Feb 2026
Cited by 1 | Viewed by 736
Abstract
The rapid progress of remote sensing (RS) and computer vision has greatly advanced change detection (CD), and hybrid architectures combining Transformers and convolutional neural networks (CNNs) have shown strong potential in recent years. Nevertheless, reliable CD for very high-resolution (VHR) imagery remains challenging [...] Read more.
The rapid progress of remote sensing (RS) and computer vision has greatly advanced change detection (CD), and hybrid architectures combining Transformers and convolutional neural networks (CNNs) have shown strong potential in recent years. Nevertheless, reliable CD for very high-resolution (VHR) imagery remains challenging due to large appearance variations across acquisition times, complex background clutter, and target structural diversity. These factors often hinder the modeling of fine edge textures, the maintenance of feature continuity, and the suppression of false changes caused by illumination fluctuations. To address these issues, this paper proposes a Cross-layer Feature Fusion Framework (CLFF) that achieves more accurate and stable change detection by explicitly enhancing the collaborative fusion capability of multi-layer features. The core component of this framework is the Multi-level Interaction Perception Block (MP-Block), which organizes effective interactions among features of different semantic levels. Based on the embedded Multi-branch Interaction Fusion Mechanism (MIFM), the MP-Block accomplishes collaborative refinement and reorganization of cross-layer features through two parallel paths for feature reconstruction and recalibration: the Response-aware Feature Reconstruction Branch (RFRB) and Adaptive Channel Group Fusion Branch (ACGF). Additionally, a lightweight position-aware attention module is introduced to adaptively modulate spatial responses, further suppressing background interference and highlighting key information related to changes. This method effectively mitigates the limitations of traditional CNNs, such as limited receptive fields and insufficient multi-layer feature interaction, while significantly enhancing the ability to collaboratively model multi-layer contextual information. To verify its effectiveness, systematic experiments were conducted on four widely used change detection benchmark datasets: LEVIR, WHU, SYSU and HRCUS. The results show that, compared to corresponding baseline models, CLFF achieves performance improvements of 1.35%, 2.78%, 3.54% and 4.85% in the IoU metric, respectively. Full article
(This article belongs to the Special Issue Remote Sensing Technology for Agricultural and Land Management)
Show Figures

Figure 1

21 pages, 4411 KB  
Article
A Methodology for Microcrack Detection in Plate Heat Exchanger Sheets Using Adaptive Templates and Features Value Analysis
by Zhibo Ding and Weiqi Yuan
Electronics 2026, 15(3), 605; https://doi.org/10.3390/electronics15030605 - 29 Jan 2026
Viewed by 501
Abstract
Aiming at the detection challenges caused by the diverse morphology of microcracks in plate heat exchanger sheets, this paper proposes a detection framework that integrates parameter-driven adaptive template generation, binary scale optimization, and feature value threshold segmentation using convolutional networks. First, based on [...] Read more.
Aiming at the detection challenges caused by the diverse morphology of microcracks in plate heat exchanger sheets, this paper proposes a detection framework that integrates parameter-driven adaptive template generation, binary scale optimization, and feature value threshold segmentation using convolutional networks. First, based on the grayscale characteristics of microcracks, an adaptive template generation model driven by key parameters (width, height, and endpoint grayscale difference) is constructed, obtaining a unique solution by solving the boundary conditions of physical features. Second, to overcome the challenge of microcrack width continuity, a binary scale optimization strategy based on the critical decay ratio k* of the correlation coefficient is designed, enabling the coverage of continuous-width defects with a finite set of templates. Finally, enhanced features are fed into a convolutional network. Utilizing the bimodal characteristic of the feature value distribution, the region corresponding to the extreme values in the top 0.3% before the foreground peak is located using 3σ extreme value statistics, achieving adaptive segmentation to identify defect regions. Evaluation on the self-built microcrack dataset SUT-B1 yielded results of 83.59% recall, 80.55% precision, and an F1 score of 81.98%. This method outperforms small object detection networks, demonstrating its advantage in morphological adaptability for small-sized objects. It also surpasses receptive field optimization modules, proving the necessity of structural optimization. The proposed method demonstrates practicality and scalability in the field of industrial inspection. Full article
Show Figures

Figure 1

27 pages, 5048 KB  
Article
MCB-RT-DETR: A Real-Time Vessel Detection Method for UAV Maritime Operations
by Fang Liu, Yongpeng Wei, Aruhan Yan, Tiezhu Cao and Xinghai Xie
Drones 2026, 10(1), 13; https://doi.org/10.3390/drones10010013 - 27 Dec 2025
Viewed by 1962
Abstract
Maritime UAV operations face challenges in real-time ship detection. Complex ocean backgrounds, drastic scale variations, and prevalent distant small targets create difficulties. We propose MCB-RT-DETR, a real-time detection transformer enhanced by multi-component boosting. This method builds upon the RT-DETR architecture. It significantly improves [...] Read more.
Maritime UAV operations face challenges in real-time ship detection. Complex ocean backgrounds, drastic scale variations, and prevalent distant small targets create difficulties. We propose MCB-RT-DETR, a real-time detection transformer enhanced by multi-component boosting. This method builds upon the RT-DETR architecture. It significantly improves detection under wave interference, lighting changes, and scale differences. Key innovations address these challenges. An Orthogonal Channel Attention (Ortho) mechanism preserves high-frequency edge details in the backbone network. Receptive Field Attention Convolution (RFAConv) enhances robustness against background clutter. A Small Object Detail Enhancement Pyramid (SOD-EPN) strengthens small-target representation. SOD-EPN combines SPDConv with multi-scale CSP-OmniKernel transformations. The neck network integrates ultra-lightweight DySample upsampling. This enables content-aware sampling for precise multi-scale localization. The method maintains high computational efficiency. Experiments on the SeaDronesSee dataset show significant improvements. MCB-RT-DETR achieves 82.9% mAP@0.5 and 49.7% mAP@0.5:0.95. These correspond to improvements of 4.5% and 3.4% relative to the baseline model. Inference speed maintains 50 FPS for real-time processing. The outstanding performance in cross-dataset tests further validates the algorithm’s strong generalization capability on DIOR remote sensing images and VisDrone2019 aerial scenes. The method provides a reliable visual perception solution for autonomous maritime UAV operations. Full article
Show Figures

Figure 1

23 pages, 11276 KB  
Article
EP-REx: Evidence-Preserving Receptive-Field Expansion for Efficient Crack Segmentation
by Sanghyuck Lee, Jeongwon Lee, Timur Khairulov, Daehyeon Kim and Jaesung Lee
Symmetry 2025, 17(10), 1653; https://doi.org/10.3390/sym17101653 - 4 Oct 2025
Viewed by 1119
Abstract
Crack segmentation plays a vital role in ensuring structural safety, yet practical deployment on resource-limited platforms demands models that balance accuracy with efficiency. While high-accuracy models often rely on computationally heavy designs to expand their receptive fields, recent lightweight approaches typically delay this [...] Read more.
Crack segmentation plays a vital role in ensuring structural safety, yet practical deployment on resource-limited platforms demands models that balance accuracy with efficiency. While high-accuracy models often rely on computationally heavy designs to expand their receptive fields, recent lightweight approaches typically delay this expansion to the deepest, low-resolution layers to maintain efficiency. This design choice leaves long-range context underutilized, where fine-grained evidence is most intact. In this paper, we propose an evidence-preserving receptive-field expansion network, which integrates a multi-scale dilated block to efficiently capture long-range context from the earliest stages and an input-guided gate that leverages grayscale conversion, average pooling, and gradient extraction to highlight crack evidence directly from raw inputs. Experiments on six benchmark datasets demonstrate that the proposed network achieves consistently higher accuracy under lightweight constraints. Each of the three proposed variants—Base, Small, and Tiny—outperforms its corresponding baselines with larger parameter counts, surpassing a total of 13 models. For example, the Base variant reduces parameters by 66% compared to the second-best CrackFormer II and floating-point operations by 53% on the Ceramic dataset, while still delivering superior accuracy. Pareto analyses further confirm that the proposed model establishes a superior accuracy–efficiency trade-off across parameters and floating-point operations. Full article
Show Figures

Figure 1

22 pages, 3632 KB  
Article
RFR-YOLO-Based Recognition Method for Dairy Cow Behavior in Farming Environments
by Congcong Li, Jialong Ma, Shifeng Cao and Leifeng Guo
Agriculture 2025, 15(18), 1952; https://doi.org/10.3390/agriculture15181952 - 15 Sep 2025
Cited by 8 | Viewed by 1939
Abstract
Cow behavior recognition constitutes a fundamental element of effective cow health monitoring and intelligent farming systems. Within large-scale cow farming environments, several critical challenges persist, including the difficulty in accurately capturing behavioral feature information, substantial variations in multi-scale features, and high inter-class similarity [...] Read more.
Cow behavior recognition constitutes a fundamental element of effective cow health monitoring and intelligent farming systems. Within large-scale cow farming environments, several critical challenges persist, including the difficulty in accurately capturing behavioral feature information, substantial variations in multi-scale features, and high inter-class similarity among different cow behaviors. To address these limitations, this study introduces an enhanced target detection algorithm for cow behavior recognition, termed RFR-YOLO, which is developed upon the YOLOv11n framework. A well-structured dataset encompassing nine distinct cow behaviors—namely, lying, standing, walking, eating, drinking, licking, grooming, estrus, and limping—is constructed, comprising a total of 13,224 labeled samples. The proposed algorithm incorporates three major technical improvements: First, an Inverted Dilated Convolution module (Region Semantic Inverted Convolution, RsiConv) is designed and seamlessly integrated with the C3K2 module to form the C3K2_Rsi module, which effectively reduces computational overhead while enhancing feature representation. Second, a Four-branch Multi-scale Dilated Attention mechanism (Four Multi-Scale Dilated Attention, FMSDA) is incorporated into the network architecture, enabling the scale-specific features to align with the corresponding receptive fields, thereby improving the model’s capacity to capture multi-scale characteristics. Third, a Reparameterized Generalized Residual Feature Pyramid Network (Reparameterized Generalized Residual-FPN, RepGRFPN) is introduced as the Neck component, allowing for the features to propagate through differentiated pathways and enabling flexible control over multi-scale feature expression, thereby facilitating efficient feature fusion and mitigating the impact of behavioral similarity. The experimental results demonstrate that RFR-YOLO achieves precision, recall, mAP50, and mAP50:95 values of 95.9%, 91.2%, 94.9%, and 85.2%, respectively, representing performance gains of 5.5%, 5%, 5.6%, and 3.5% over the baseline model. Despite a marginal increase in computational complexity of 1.4G, the algorithm retains a high detection speed of 147.6 frames per second. The proposed RFR-YOLO algorithm significantly improves the accuracy and robustness of target detection in group cow farming scenarios. Full article
(This article belongs to the Section Farm Animal Production)
Show Figures

Figure 1

17 pages, 2482 KB  
Article
Coastline Identification with ASSA-Resnet Based Segmentation for Marine Navigation
by Yuhan Wang, Weixian Li, Zhengxun Zhou and Ning Wu
Appl. Sci. 2025, 15(16), 9113; https://doi.org/10.3390/app15169113 - 19 Aug 2025
Cited by 3 | Viewed by 1022
Abstract
Real-time and accurate segmentation of coastlines is of paramount importance for the safe navigation of unmanned surface vessels (USVs). Classical methods such as U-Net and DeepLabV3 have been proven to be effective in coastline segmentation tasks. However, their performance substantially degrades in real-world [...] Read more.
Real-time and accurate segmentation of coastlines is of paramount importance for the safe navigation of unmanned surface vessels (USVs). Classical methods such as U-Net and DeepLabV3 have been proven to be effective in coastline segmentation tasks. However, their performance substantially degrades in real-world scenarios due to variations in lighting and environmental conditions, particularly from water surface reflections. This paper proposes an enhanced ResNet-50 model, namely ASSA-ResNet, for coastline segmentation for vision-based marine navigation. ASSA-ResNet integrates Atrous Spatial Pyramid Pooling (ASPP) to expand the model’s receptive field and incorporates a Global Channel Spatial Attention (GCSA) module to suppress interference from water reflections. Through feature pyramid fusion, ASSA-ResNet reinforces the semantic representation of features at various scales to ensure precise boundary delineation. The performance of ASSA-ResNet is validated with a dataset encompassing diverse brightness conditions and scenarios. Notably, mean Pixel Accuracy (mPA) and mean Intersection over Union (mIoU) of 98.90% and 98.17%, respectively, have been achieved on the self-constructed dataset, with corresponding values of 99.18% and 98.39% observed on the USVInland unmanned vessel dataset. Comparative analyses reveal that ASSA-ResNet outperforms the U-Net model by 1.78% in mPA and 2.9% in mIOU relative to the DeepLabV3 model. It also demonstrates enhancements of 1.85% in mPA and 3.19% in mIoU. On the USVInland dataset, ASSA-ResNet exhibits superior performance compared to U-Net, with improvements of 0.41% in mPA and 0.12% in mIoU, while surpassing DeepLabV3 by 0.33% in mPA and 0.21% in mIoU. Full article
(This article belongs to the Section Marine Science and Engineering)
Show Figures

Figure 1

17 pages, 7350 KB  
Article
Lightweight Network for Spoof Fingerprint Detection by Attention-Aggregated Receptive Field-Wise Feature
by Md Al Amin, Naim Reza and Ho Yub Jung
Electronics 2025, 14(9), 1823; https://doi.org/10.3390/electronics14091823 - 29 Apr 2025
Cited by 2 | Viewed by 3364
Abstract
The spread of biometric systems utilizing fingerprints has increased the need for advanced spoof detection techniques, but training convolutional neural networks (CNNs) with the limited number of images available in fingerprint datasets poses significant challenges. In this paper, we propose a lightweight network [...] Read more.
The spread of biometric systems utilizing fingerprints has increased the need for advanced spoof detection techniques, but training convolutional neural networks (CNNs) with the limited number of images available in fingerprint datasets poses significant challenges. In this paper, we propose a lightweight network architecture which addresses the challenges inherent in small fingerprint datasets by employing a moderately deep network architecture which is sufficient for extracting essential features from fingerprint images. We apply a hyperbolic tangent activation to the final feature map, which has features from local receptive fields, and average the responses into a single value. Thus, our architecture reduces overfitting by increasing the number of effective labels during training. Additionally, the incorporation of the spatial attention module enhances feature representation, culminating in improved accuracy. The evaluation results show that the proposed model, with only 0.14 million parameters, outperforms existing techniques including lightweight models and transfer-learning-based models, achieving superior average test accuracies of 98.30% and 95.57% on the LivDet-2015 and -2017 datasets, respectively. It also delivers state-of-the-art cross-material performance, with corresponding average classification error values of 0.81% and 1.91%, making it highly effective for on-device fingerprint authentication. Full article
Show Figures

Figure 1

Back to TopTop