Search Results (809)

Current agricultural irrigation management practices are often extensive, and traditional soil moisture content (SMC) monitoring methods are inefficient, so there is a pressing need for innovative approaches in precision irrigation. This study proposes a Multi-Feature Fusion Network (MFF-Net) for SMC inversion. The model uses a designed Channel-Changeable Residual Block (ResBlockCC) to construct a multi-branch feature extraction and fusion architecture. Integrating the Channel Squeeze and Spatial Excitation (sSE) attention module with U-Net-like skip connections, MFF-Net inverts root-zone SMC from summer maize leaf images. Field experiments were conducted in Zhengzhou, Henan Province, China, from 2024 to 2025, under three irrigation treatments: 60–70% θ_fc, 70–90% θ_fc, and 60–90% θ_fc (θ_fc denotes field capacity). This study shows that (1) MFF-Net achieved its smallest inversion error under the 60–70% θ_fc treatment, suggesting the inversion was most effective when SMC variation was small and relatively low; (2) MFF-Net demonstrated superior performance to several benchmark models, achieving an R² of 0.84; and (3) the ablation study confirmed that each feature branch and the sSE attention module contributed positively to model performance. MFF-Net thus offers a technological reference for real-time precision irrigation and shows promise for field SMC inversion in summer maize. Full article

Intracardiac echocardiography (ICE) is increasingly recognized as a valuable imaging modality in structural heart interventions, offering high-resolution, real-time visualization from within the cardiac chambers. Originally developed for electrophysiologic procedures, ICE has expanded its use across a broad spectrum of structural interventions, including atrial septal defect (ASD) and patent foramen ovale (PFO) closure, left atrial appendage occlusion (LAAO), transseptal puncture guidance, transcatheter edge-to-edge repair (TEER), balloon mitral valvuloplasty, and both mitral and tricuspid valve therapies. This review outlines the current role and technical principles of ICE, with an emphasis on catheter design, image acquisition protocols, and the emerging potential of 3D ICE. Comparisons with transesophageal echocardiography (TEE) and fluoroscopy are discussed, highlighting ICE’s ability to support minimally invasive, sedation-sparing procedures while maintaining procedural precision. We provide a focused analysis of ICE-guided applications in specific clinical scenarios, emphasizing its role in anatomical assessment, device navigation, and intra-procedural monitoring. Data from recent clinical studies and registries are reviewed to assess safety, feasibility, and outcomes. Practical considerations including operator learning curve, workflow integration, and limitations such as cost and field of view are also addressed. Lastly, we explore future directions including advanced 3D imaging, fusion imaging, artificial intelligence integration, and robotic catheter systems. Full article

While deep learning methods are increasingly applied in the field of DOA estimation, existing approaches generally feed the real and imaginary parts of the covariance matrix directly into neural networks without optimizing the input features, which prevents classical attention mechanisms from improving accuracy. This paper proposes a spatio-temporal fusion model named DACL-Net for DOA estimation. The spatial branch applies a two-dimensional Fourier transform (2D-FT) to the covariance matrix, causing angles to appear as peaks in the magnitude spectrum. This operation transforms the original covariance matrix into a dark image with bright spots, enabling the convolutional neural network (CNN) to focus on the bright-spot components via an attention module. Additionally, a spectrum attention mechanism (SAM) is introduced to enhance the extraction of temporal features in the time branch. The model learns simultaneously from two data branches and finally outputs DOA results through a linear layer. Simulation results demonstrate that DACL-Net outperforms existing algorithms in terms of accuracy, achieving an RMSE of 0.04${}^{\circ }$ at an SNR of 0 dB. Full article

Magnetic resonance imaging (MRI) super-resolution (SR) enables high-resolution reconstruction from low-resolution acquisitions, reducing scan time and easing hardware demands. However, most deep learning-based SR models are large and computationally heavy, limiting deployment in clinical workstations, real-time pipelines, and resource-restricted platforms such as low-field and portable MRI. We introduce CHARMS, a lightweight convolutional–Transformer hybrid with attention regularization optimized for MRI SR. CHARMS employs a Reverse Residual Attention Fusion backbone for hierarchical local feature extraction, Pixel–Channel and Enhanced Spatial Attention for fine-grained feature calibration, and a Multi-Depthwise Dilated Transformer Attention block for efficient long-range dependency modeling. Novel attention regularization suppresses redundant activations, stabilizes training, and enhances generalization across contrasts and field strengths. Across IXI, Human Connectome Project Young Adult, and paired 3T/7T datasets, CHARMS (~1.9M parameters; ~30 GFLOPs for 256 × 256) surpasses leading lightweight and hybrid baselines (EDSR, PAN, W2AMSN-S, and FMEN) by 0.1–0.6 dB PSNR and up to 1% SSIM at ×2/×4 upscaling, while reducing inference time ~40%. Cross-field fine-tuning yields 7T-like reconstructions from 3T inputs with ~6 dB PSNR and 0.12 SSIM gains over native 3T. With near-real-time performance (~11 ms/slice, ~1.6–1.9 s per 3D volume on RTX 4090), CHARMS offers a compelling fidelity–efficiency balance for clinical workflows, accelerated protocols, and portable MRI. Full article

Background/Objectives: Despite significant advances in imaging technology, real-time intra-fraction monitoring of moving targets remains a challenge in markerless radiotherapy. This retrospective study investigates the use of ExacTrac Dynamic by Brainlab as an intra-fraction monitoring tool for stereotactic body radiotherapy (SBRT) in both early-stage NSCLC and oligometastatic disease. Methods: A total of 63 X-ray pairs from 21 patients were analyzed to evaluate tumor visualization with and without a surrogate approach. Statistical analysis was conducted to determine whether failures could be attributed to tumor size or localization using the Mann–Whitney U-test and Fisher’s exact test. The accuracy of the X-ray/digitally reconstructed radiograph (DRR) surrogate-based fusion was assessed by calculating and comparing the corresponding 3D vectors according to the linear mixed effects model, with a random slope effect for size of surrogate and a random intercept per patient. Results: Surrogates enhanced tumor visualization on X-ray/DRR fusions from 14.3% to 75.5%. Tumor size and lung affected (left or right) did not predict visualization success. Tumor location, however, tended to influence visibility, with lesions in the upper lobes being more readily visualized (88%) than those in the lower lobes (48.1%), although no statistical significance was reported (p > 0.05). Regarding geometric accuracy, 76% of the analyzed data points deviated less than 5 mm in the 3D vector measurements, the mean values were around 4 mm (±3 mm), and the medians were within 3 mm across all conditions. No statistically significant differences (p > 0.05) were found based on the surrogate size or the triggering time of the X-ray during the breathing cycle. Conclusions: Surrogate-based DRRs, referred to as Correlation Objects, demonstrate consistent geometric accuracy across multiple surrogate sizes and X-ray acquisitions, supporting the clinical translation of markerless lung targeting workflows for lung SBRT. Full article

In the context of bearing anomaly detection, challenges such as imbalanced sample distribution and complex operational conditions present significant difficulties for data-driven deep learning models. These issues often result in overfitting and high false positive rates in complex real-world scenarios. This paper proposes a strategy that leverages multimodal fusion and Self-Adversarial Training (SAT) to construct and train a deep learning model. First, the one-dimensional bearing vibration time-series data are converted into Gramian Angular Difference Field (GADF) images, and multimodal feature fusion is performed with the original time-series data to capture richer spatiotemporal correlation features. Second, a composite data augmentation strategy combining time-domain and image-domain transformations is employed to effectively expand the anomaly samples, mitigating data scarcity and class imbalance. Finally, the SAT mechanism is introduced, where adversarial samples are generated within the fused feature space to compel the model to learn more generalized and robust feature representations, thereby significantly enhancing its performance in realistic and noisy environments. Experimental results demonstrate that the proposed method outperforms traditional baseline models across key metrics such as accuracy, precision, recall, and F1-score in abnormal bearing anomaly detection. It exhibits exceptional robustness against rail-specific interferences, offering a specialized solution strictly tailored for the unique, high-noise operational environments of intelligent railway maintenance. Full article

Optical and SAR image registration and fusion are pivotal in the remote sensing field, as they leverage the complementary advantages of both modalities. However, achieving this with high accuracy and efficiency remains challenging. This challenge arises because traditional methods are confined to the image domain, applied after independent image formation. They attempt to correct geometric mismatches that are rooted in fundamental physical differences, an approach that inherently struggles to achieve both precision and speed. Therefore, this paper introduces a co-designed system and algorithm framework to overcome the fundamental challenges. At the system level, we pioneer an innovative airborne co-aperture system to ensure synchronous data acquisition. At the algorithmic level, we derive a theoretical model within the wavenumber domain imaging process, attributing optical/SAR pixel deviations to the deterministic phase errors introduced by its core Stolt interpolation operation. This model enables a signal-domain compensation technique, which employs high-order Fourier series fitting to correct these errors during the SAR image formation itself. This co-design yields a unified processing pipeline that achieves direct, sub-pixel co-registration, thereby establishing a foundational paradigm for real-time multi-source data processing. The experimental results on both multi-point and structural targets confirm that our method achieves sub-pixel registration accuracy across diverse scenarios, accompanied by a marked gain in computational efficiency over the time-domain approach. Full article

With the explosive growth in the quantity of electrical equipment in distribution networks, traditional manual inspection struggles to achieve comprehensive coverage due to limited manpower and low efficiency. This has led to frequent equipment failures including partial discharge, insulation aging, and poor contact. These issues seriously compromise the safe and stable operation of distribution networks. Real-time monitoring and defect identification of their operation status are critical to ensuring the safety and stability of power systems. Currently, commonly used methods for defect identification in distribution network electrical equipment mainly rely on single-image or voiceprint data features. These methods lack consideration of the complementarity and interleaved nature between image and voiceprint features, resulting in reduced identification accuracy and reliability. To address the limitations of existing methods, this paper proposes distribution network electrical equipment defect identification based on multi-modal image voiceprint data fusion and channel interleaving. First, image and voiceprint feature models are constructed using two-dimensional principal component analysis (2DPCA) and the Mel scale, respectively. Multi-modal feature fusion is achieved using an improved transformer model that integrates intra-domain self-attention units and an inter-domain cross-attention mechanism. Second, an image and voiceprint multi-channel interleaving model is applied. It combines channel adaptability and confidence to dynamically adjust weights and generates defect identification results using a weighting approach based on output probability information content. Finally, simulation results show that, under the dataset size of 3300 samples, the proposed algorithm achieves a 8.96–33.27% improvement in defect recognition accuracy compared with baseline algorithms, and maintains an accuracy of over 86.5% even under 20% random noise interference by using improved transformer and multi-channel interleaving mechanism, verifying its advantages in accuracy and noise robustness. Full article

Conventional High Dynamic Range (HDR) image fusion algorithms generally require two or more original images with different exposure times for synthesis, making them unsuitable for real-time processing scenarios such as video streams. Additionally, the synthesized HDR images have the same bit depth as the original images, which may lead to banding artifacts and limits their applicability in professional fields requiring high fidelity. This paper utilizes a Field Programmable Gate Array (FPGA) to support an image sensor operating in Clear HDR mode, which simultaneously outputs High Conversion Gain (HCG) and Low Conversion Gain (LCG) images. These two images share the same exposure duration and are captured at the same moment, making them well-suited for real-time HDR fusion. This approach provides a feasible solution for real-time processing of video streams. An adaptive adjustment algorithm is employed to address the requirement for high fidelity. First, the initial HCG and LCG images are fused under the initial fusion parameters to generate a preliminary HDR image. Subsequently, the gain of the high-gain images in the video stream is adaptively adjusted according to the brightness of the fused HDR image, enabling stable brightness under dynamic illumination conditions. Finally, by evaluating the read noise of the HCG and LCG images, the fusion parameters are adaptively optimized to synthesize an HDR image with higher bit depth. Experimental results demonstrate that the proposed method achieves a processing rate of 46 frames per second for 2688 × 1520 resolution video streams, enabling real-time processing. The bit depth of the image is enhanced from 12 bits to 16 bits, preserving more scene information and effectively addressing banding artifacts in HDR images. This improvement provides greater flexibility for subsequent image processing tasks. Consequently, the adaptive algorithm is particularly suitable for dynamically changing scenarios such as real-time surveillance and professional applications including industrial inspection. Full article

Real-time monitoring of hazardous gas emissions in open environments remains a critical challenge. Conventional spectrometers and filter wheel systems acquire spectral and spatial information sequentially, which limits their ability to capture multiple gas species and dynamic dispersion patterns rapidly. A High-Resolution Snapshot Multispectral Imaging System (HRSMIS) is proposed to integrate high spatial fidelity with multispectral capability for near real-time plume visualization, gas species identification, and concentration retrieval. Operating across the 7–14 $\mathsf{\mu }$m spectral range, the system employs a dual-path optical configuration in which a high-resolution imaging path and a multispectral snapshot path share a common telescope, allowing for the simultaneous acquisition of fine two-dimensional spatial morphology and comprehensive spectral fingerprint information. Within the multispectral path, two $5\times 5$ microlens arrays (MLAs) combined with a corresponding narrowband filter array generate 25 distinct spectral channels, allowing concurrent detection of up to 25 gas species in a single snapshot. The high-resolution imaging path provides detailed spatial information, facilitating spatio-spectral super-resolution fusion for multispectral data without complex image registration. The HRSMIS demonstrates modulation transfer function (MTF) values of at least 0.40 in the high-resolution channel and 0.29 in the multispectral channel. Monte Carlo tolerance analysis confirms imaging stability, enabling the real-time visualization of gas plumes and the accurate quantification of dispersion dynamics and temporal concentration variations. Full article

Cotton is an important economic crop, and its weed management directly affects yield and quality. In actual cotton fields, detection accuracy still faces challenges due to the complex types of weeds, variable morphologies, and environmental factors. Most existing models rely on the attention mechanism to improve performance, but channel attention tends to ignore spatial information, while full spatial attention brings high computational costs. Therefore, this paper proposes a grouped enhanced fusion attention mechanism (GEFA), which combines grouped convolution and local spatial attention to reduce complexity and parameter quantity while effectively enhancing feature expression ability. The GEFAY detection model constructed based on GEFA achieves good balance in efficiency, accuracy, and complexity on the CottonWeedDet12, VOC, and COCO datasets. Compared with classic attention methods, this model has the smallest increase in parameters and computational costs while significantly improving accuracy. It is more suitable for deployment on edge devices. The further designed end-to-end intelligent weed detection system and edge device deployment can achieve image detection on local maps and real-time cameras, with good practicality and scalability, providing effective technical support for intelligent visual applications in precision agriculture. Full article

Cassava remains a critical food-security crop across Africa and Southeast Asia but is highly vulnerable to diseases such as cassava mosaic disease (CMD) and cassava brown streak disease (CBSD). Traditional diagnostic approaches are slow, labor-intensive, and inconsistent under field conditions. This review synthesizes current advances in combining unmanned aerial vehicles (UAVs) with deep learning (DL) to enable scalable, data-driven cassava disease detection. It examines UAV platforms, sensor technologies, flight protocols, image preprocessing pipelines, DL architectures, and existing datasets, and it evaluates how these components interact within UAV–DL disease-monitoring frameworks. The review also compares model performance across convolutional neural network-based and Transformer-based architectures, highlighting metrics such as accuracy, recall, F1-score, inference speed, and deployment feasibility. Persistent challenges—such as limited UAV-acquired datasets, annotation inconsistencies, geographic model bias, and inadequate real-time deployment—are identified and discussed. Finally, the paper proposes a structured research agenda including lightweight edge-deployable models, UAV-ready benchmarking protocols, and multimodal data fusion. This review provides a consolidated reference for researchers and practitioners seeking to develop practical and scalable cassava-disease detection systems. Full article

Ship detection serves as a core foundational task for marine environmental perception. However, in real marine scenarios, dense vessel traffic often causes severe target occlusion while multi-scale targets, asymmetric vessel geometries, and harsh conditions (e.g., haze, low illumination) further degrade image quality. These factors pose significant challenges to vision-based ship detection methods. To address these issues, we propose ${\mathrm{M}}^2\mathrm{S}\text{-}\mathrm{YOLOv}8$, an improved framework based on YOLOv8, which integrates three key enhancements: First, a Multi-Scale Asymmetry-aware Parallelized Patch-wise Attention (MSA-PPA) module is designed in the backbone to strengthen the perception of multi-scale and geometrically asymmetric vessel targets. Second, a Deformable Convolutional Upsampling (DCNUpsample) operator is introduced in the Neck network to enable adaptive feature fusion with high computational efficiency. Third, a Wasserstein-Distance-Based Weighted Normalized CIoU (WA-CIoU) loss function is developed to alleviate gradient imbalance in small-target regression, thereby improving localization stability. Experimental results on the Unmanned Vessel Zhoushan Perception Dataset (UZPD) and the open-source Singapore Maritime Dataset (SMD) demonstrate that ${\mathrm{M}}^2\mathrm{S}\text{-}\mathrm{YOLOv}8$ achieves a balanced performance between lightweight design and real-time inference, showcasing strong potential for reliable deployment on edge devices of unmanned marine platforms. Full article

In the field management of maize, phenomena such as missed sowing and empty seedlings directly affect the final yield. By implementing seedling replenishment activities and promptly evaluating seedling growth, maize output can be increased by improving seedling survival rates. To address the challenges posed by complex field environments (including varying light conditions, weeds, and foreign objects), as well as the performance limitations of model deployment on resource-constrained devices, this study proposes a Lightweight Real-Time You Only Look Once (LRT-YOLO) model. This model builds upon the You Only Look Once version 11n (YOLOv11n) framework by designing a lightweight, optimized feature architecture (OF) that enables the model to focus on the characteristics of small to medium-sized maize seedlings. The feature fusion network incorporates two key modules: the Feature Complementary Mapping Module (FCM) and the Multi-Kernel Perception Module (MKP). The FCM captures global features of maize seedlings through multi-scale interactive learning, while the MKP enhances the network’s ability to learn multi-scale features by combining different convolution kernels with pointwise convolution. In the detection head component, the introduction of an NMS-free design philosophy has significantly enhanced the model’s detection performance while simultaneously reducing its inference time. The experiments show that the $mA{P}_{50}$ and $mA{P}_{50:95}$ of the LRT-YOLO model reached 95.9% and 63.6%, respectively. The model has only 0.86M parameters and a size of just 2.35 M, representing reductions of 66.67% and 54.89% in the number of parameters and model size compared to YOLOv11n. To enable mobile deployment in field environments, this study integrates the LRT-YOLO model with the ByteTrack multi-object tracking algorithm and deploys it on the NVIDIA Jetson AGX Orin platform, utilizing OpenCV tools to achieve real-time visualization of maize seedling tracking and counting. Experiments demonstrate that the frame rate (FPS) achieved with TensorRT acceleration reached 23.49, while the inference time decreased by 38.93%. Regarding counting performance, when tested using static image data, the coefficient of determination ($R^2$) and root mean square error (RMSE) were 0.988 and 5.874, respectively. The cross-line counting method was applied to test the video data, resulting in an $R^2$ of 0.971 and an RMSE of 16.912, respectively. Experimental results show that the proposed method demonstrates efficient performance on edge devices, providing robust technical support for the rapid, non-destructive counting of maize seedlings in field environments. Full article

Non-destructive prediction of harvest timing is increasingly important in greenhouse melon cultivation, yet image-based methods alone often fail to reflect environmental factors affecting fruit development. Likewise, environmental or fertigation data alone cannot capture fruit-level variation. This gap calls for a multimodal approach integrating both sources of information. This study presents a fusion model combining RGB images with environmental and fertigation data to predict optimal harvest timing for melons. A YOLOv8n-based model detected fruits and estimated diameters under marker and no-marker conditions, while an LSTM processed time-series variables including temperature, humidity, CO₂, light intensity, irrigation, and electrical conductivity. The extracted features were fused through a late-fusion strategy, followed by an MLP for predicting diameter, biomass, and harvest date. The marker condition improved detection accuracy; however, the no-marker condition also achieved sufficiently high performance for field application. Diameter and weight showed a strong correlation (R² > 0.9), and the fusion model accurately predicted the actual harvest date of 28 August 2025. These results demonstrate the practicality of multimodal fusion for reliable, non-destructive harvest prediction and highlight its potential to bridge the gap between controlled experiments and real-world smart farming environments. Full article