Search Results (202)

Hyperspectral image (HSI) classification regarding vegetation is hampered by strong intra-class spectral variability and inter-class similarity, and commonly used random pixel splits can introduce spatial-context leakage that inflates test accuracy in patch-based models. To address these issues, we propose a classification framework that couples a leakage-free block partition (LFBP) strategy with class-aware multi-prototype contrastive loss (CAMP-CL). LFBP assigns non-overlapping spatial blocks to training/validation/test sets and reserves a buffer matched to the patch radius to prevent contextual overlap while keeping class distributions balanced. CAMP-CL represents each class with multiple learnable prototypes and performs supervised contrastive learning at the prototype level, encouraging compact yet multimodal intra-class embedding and improved inter-class separation. Experiments conducted on the Matiwan Village airborne HSI dataset under the LFBP protocol show that the proposed method can achieve 91.51% overall accuracy (OA) and 91.49% average accuracy (AA). Compared with the strongest baseline, supervised contrastive learning (SupCon), the proposed method yields consistent gains of 1.07 percentage points (pp) in both OA and AA while improving OA by 5.76 pp over the cross-entropy baseline. The results suggest that CAMP-CL is beneficial for addressing the challenges of HSI classification for fine-grained vegetation, while leakage-free evaluation protocols are important for obtaining more reliable performance estimates in practical settings. Full article

Customizing diffusion models via Low-Rank Adaptation (LoRA) has become a standard approach for customized concept injection. However, synthesizing multiple customized concepts within a single image remains challenging due to the parameter pollution problem, where naive fusion leads to gradient conflicts and severe quality degradation. In this paper, we introduce ConWave-LoRA, a novel framework designed to achieve hierarchical disentanglement of object and style concepts in LoRAs. Supported by our empirical validation regarding frequency distribution in the latent space, we identify that object identities are predominantly encoded in high-frequency structural perturbations, while artistic styles manifest through low-frequency global layouts. Leveraging this insight, we propose a Discrete Wavelet Transform (DWT) based filtering strategy that projects these concepts into orthogonal optimization subspaces during contrastive learning, thereby isolating structural details from stylistic attributes. Extensive experiments, including expanded ablation studies on LoRA rank sensitivity and style consistency, demonstrate that ConWave-LoRA consistently outperforms strong baselines, producing high-fidelity images that successfully integrate multiple distinct concepts without interference. Full article

Two folding screens by futurist artist Giacomo Balla (1871–1958) in the collection of the Kröller-Müller Museum (the Netherlands) were investigated: Paravento con linea di velocità (1916–1917) and Paravento (1916/1917–1958). The screens are painted on both sides, the first on four canvases, stretched onto two wooden strainers and framed with painted wooden strips, and the second on wooden panels set into four painted stiles. In the past, damages on Paravento con linea di velocità were restored by conservators, while Paravento was probably first reworked by the artist himself and later restored by conservators. Yellowed varnish and discolored retouches on both screens led to a wish for treatment. The aim of this research was to gain insight into the painting techniques, layer buildup, pigments, binders, and varnishes of the two artworks. This information supported the decision making for treatment, and it broadens the knowledge on the materials used by Balla. Up to now, only a few published studies deal with the technical examination of paintings by this artist. Both folding screens were subjected to technical photography (UV, IR photography, and X-ray) and were examined with portable point X-ray fluorescence (pXRF) and Raman spectroscopy. Moreover, samples were taken. Cross-sections were studied with optical microscopy, scanning electron microscopy with energy-dispersive X-ray spectroscopy (SEM-EDX), attenuated total reflection Fourier-transform infrared (ATR-FTIR) imaging, and micro-Raman spectroscopy. Loose samples were examined with SEM-EDX, FTIR and micro-Raman spectroscopy, and pyrolysis gas chromatography mass spectrometry (Py-GC/MS). For Paravento con linea di velocità, all pigments and fillers of the painted canvases are compatible with the dating of the screen (1916–1917), but they differ from those on the frame. Here, rutile, in combination with various pigments, among which are blue copper phthalocyanine (PB15) and other synthetic organic pigments, was found. This indicates that the frame has been painted later, likely after the Second World War. The composition of the binders differs as well. Drying oil and pine resin have been used on the canvases, explaining the smooth and glossy appearance and solvent-sensitivity of the paint. On the frame, oil with some alkyd resin was identified. The provenance of the screen before 1972 is not clear, nor when the frame was made and painted and by whom. The results for Paravento indicate that the palettes of the two sides—painted in different styles—are comparable. Mainly inorganic pigments were found, except for the dark red areas, where toluidine red (PR3) is present. pXRF showed high amounts of zinc; cross-sections revealed that zinc white is present in the lower layers. These pigments are compatible with the dating of the screen (1916–1917). In many of the upper paint layers though, except for some green, dark red, and black areas, rutile has been identified. This indicates that these layers were applied later, likely after the Second World War. Since this folding screen was used by the artist and his family until his death in 1958, it seems likely that Balla himself reworked the screen. Full article

Synthetic aperture radar (SAR) is an active microwave imaging system equipped with penetration capability, enabling all-time and all-weather Earth observation, and demonstrates significant advantages in large-scale surface water-body detection. Although SAR images can provide relatively clear water-body details, they are susceptible to interference from external factors such as complex terrain and background noise, resulting in fragmented detection outcomes and poor connectivity. Therefore, a Connectivity Refinement Network (ConRNet) is proposed in this study to address the issue of fragmented water-body regions in water-body detection results, combining HISEA-1 and Chaohu-1 SAR data. ConRNet is equipped with attention mechanisms and a connectivity prediction module, combined with dual supervision from segmentation and connectivity labels. Unlike conventional attention modules that only emphasize pixel-wise saliency, the proposed Dual Self-Attention Module (DSAM) jointly captures spatial and channel dependencies. Meanwhile, the Connectivity Prediction Module (CPM) reformulates water-body connectivity as a regression problem to directly optimize structural coherence without relying on post-processing. Leveraging dual supervision from segmentation and connectivity labels, ConRNet achieves simultaneous improvements in topological consistency and pixel-level accuracy. The performance of the proposed ConRNet is evaluated by con-ducting comparative experiments with five deep learning models: FCN, U-Net, DeepLabv3+, HRNet, and MAGNet. The experimental results demonstrate that the ConRNet achieves the highest accuracy in water-body detection, with an intersection over union (IoU) of 88.59% and an F1-score of 93.87%. Full article

Few-Shot Class-Incremental Learning (FSCIL) aims to continually learn novel classes from limited data while retaining knowledge of previously learned classes. To mitigate catastrophic forgetting, most approaches pre-train a powerful backbone on the base session and keep it frozen during incremental sessions. Within this framework, existing studies primarily focus on representation learning in FSCIL, particularly Self-Supervised Contrastive Learning (SSCL), to enhance the transferability of representations and thereby boost model generalization to novel classes. However, they face a trade-off dilemma: improving transferability comes at the expense of discriminability, precluding simultaneous high performance on both base and novel classes. To address this issue, we propose BR-FSCIL, a representation learning framework for the FSCIL scenario. In the pre-training stage, we first design a Hierarchical Contrastive Learning (HierCon) algorithm. HierCon leverages label information to model hierarchical relationships among features. In contrast to SSCL, it maintains strong discriminability when promoting transferability. Second, to further improve the model’s performance on novel classes, an Alignment Modulation (AM) loss is proposed that explicitly facilitates learning of knowledge shared across classes from an inter-class perspective. Building upon the hierarchical discriminative structure established by HierCon, it additionally improves the model’s adaptability to novel classes. Through optimization at both intra-class and inter-class levels, the representations learned by BR-FSCIL achieve a balance between discriminability and transferability. Extensive experiments on mini-ImageNet, CIFAR100, and CUB200 demonstrate the effectiveness of our method, which achieves final session accuracies of 53.83%, 53.04%, and 62.60%, respectively. Full article

Due to the growing complexity of video encoders, the optimization of the parameters of the encoding process is becoming an important issue. In recent years, this has become an important field of application of neural networks. Artificial neural networks in video encoders are used to accelerate the video encoder operation. This paper demonstrates the use of different ResNet- and DenseNet-type architectures to accelerate the CTU partitioning algorithm in HEVC in All Intra mode. The paper demonstrates the results of an exhaustive evaluation of different proposed architectures, considering compression efficiency, network size, and encoding time reduction. Multiple pros and cons of the proposed architectures are presented in the Conclusions, considering various limitations that may be important for a given application, like hardware-constrained sensor networks or standalone small devices operating with images and videos. Full article

Background/Objectives: Breast cancer (BrCA) is among the deadliest cancers for women in the world. The disease has four distinct molecular subtypes which can be determined by gene expression profiling. Understanding these subtypes has enabled the development of targeted therapeutics. Additionally, following initial successful treatment, some patients experience disease recurrence events. Methods: In this study, we used radiomics coupled with machine learning techniques to predict molecular subtypes and disease recurrence events from a dataset of MRI features deriving from a single-institutional, retrospective collection of 922 biopsy-confirmed invasive BrCA patients. The feature-rich and comprehensive dataset consists of radiomic as well as demographic, clinical, and molecular subtype information. We focused our analyses on Black and White patients who were 50 years or younger at diagnosis (n = 346) to identify racial disparities that exist between molecular subtypes and disease recurrence events. Random Forest and AdaBoostM1 were applied to over 500 radiomics features. Results: Radiomics alone or combined with gene expression data can accurately predict molecular subtype and disease recurrence events for both racial groups. In total, we found over 40 radiomics features that have significant associations with race. The radiomic features that are most predictive in the Breast and Fibroglandular Tissue Volume imaging category for Black patients was breast volume (Breast_Vol) and for White patients was post contrast tissue volume (TissueVol_PostCon). Conclusions: These results suggest that radiomics can be used to predict differences in BrCA recurrence and molecular subtype between racial groups and can have an impact on clinical outcomes. Full article

Dental implant components are typically fabricated using subtractive manufacturing, often involving metal materials that can be costly, inefficient, and time-consuming. This study explores the use of additive manufacturing (AM) with zirconia for dental implant overdenture bars, focusing on mechanical performance, stress distribution, and fit. Solid and lattice-structured bars were designed in Fusion 360 and produced using LithaCon 210 3Y-TZP zirconia (Lithoz GmbH, Vienna, Austria) on a CeraFab 8500 printer. Post-processing included cleaning, debinding, and sintering. A 3D-printed denture was also fabricated to evaluate fit. Thermography and optical imaging were used to assess adaptation. Custom fixtures were developed for flexural testing, and fracture loads were recorded to calculate stress distribution using finite element analysis (ANSYS R2025). The FEA model assumed isotropic, homogeneous, linear-elastic material behavior. Bars were torqued to 15 Ncm on implant analogs. The average fracture loads were 1.2240 kN (solid, n = 12) and 1.1132 kN (lattice, n = 5), with corresponding stress values of 147 MPa and 143 MPa, respectively. No statistically significant difference was observed (p = 0.578; α = 0.05). The fracture occurred near high-stress regions at fixture support points. All bars demonstrated a clinically acceptable fit on the model; however, further validation and clinical evaluation are still needed. Additively manufactured zirconia bars, including lattice structures, show promise as alternatives to conventional superstructures, potentially offering reduced material use and faster production without compromising mechanical performance. Full article

The classification of submarine topography and geomorphology is essential for marine resource exploitation and ocean engineering, with wide-ranging implications in marine geology, disaster assessment, resource exploration, and autonomous underwater navigation. Submarine landscapes are highly complex and diverse. Traditional visual interpretation methods are not only inefficient and subjective but also lack the precision required for high-accuracy classification. While many machine learning and deep learning models have achieved promising results in image classification, limited work has been performed on integrating backscatter and bathymetric data for multi-source processing. Existing approaches often suffer from high computational costs and excessive hyperparameter demands. In this study, we propose a novel approach that integrates pruning-enhanced ConDenseNet with label smoothing regularization to reduce misclassification, strengthen the cross-entropy loss function, and significantly lower model complexity. Our method improves classification accuracy by 2% to 10%, reduces the number of hyperparameters by 50% to 96%, and cuts computation time by 50% to 85.5% compared to state-of-the-art models, including AlexNet, VGG, ResNet, and Vision Transformer. These results demonstrate the effectiveness and efficiency of our model for multi-source submarine topography classification. Full article

The lemon flavonoid extract Eriomin^® (LE), which is rich in eriocitrin, has demonstrated antioxidant and anti-inflammatory properties in both animal and human studies. Given the established interplay among aging, oxidative stress, and inflammation, this study investigated the influences of LE on the pituitary–adrenal (PA) axis in aged rats and its potential to mitigate age-related physiological changes in this system. The effects of LE (40 mg/kg/day suspended in sunflower oil) on the morphofunctional properties of the PA axis were studied in 24-month-old male Wistar rats following four weeks of oral treatment. Control groups included vehicle-treated (sunflower oil; CON) and untreated intact controls (ICON). Stereological and imaging analyses revealed no significant changes in pituitary ACTH cells; however, Pomc gene expression was significantly downregulated in the LE group compared to both controls (p ≤ 0.05). LE treatment resulted in a significant reduction in adrenal gland weight (p ≤ 0.05), adrenal gland volume (p ≤ 0.01), zona fasciculata (ZF) volume (p ≤ 0.01) and ZF cell volume (p ≤ 0.05). These changes were accompanied by a significant decrease in serum corticosterone levels (p ≤ 0.05). In conclusion, LE downregulated PA axis activity in aged rats. Considering the association between age-related increases in PA activity and adverse health outcomes, citrus flavonoid extracts such as LE may hold promise as anti-aging supplements aimed at mitigating age-related stress dysregulation. Full article

The paper makes a novel contribution to bridging the academic–practitioner divide in tourism studies, specifically in the context of the destination image. An advanced, robust estimation modeling approach that analyzed a lagged commercial international survey of potential tourists reveals that academics and practitioners tend to draw different conclusions from the same dataset based on their different hypotheses. These findings suggest that academics and practitioners have limited perspectives of destination image, casting doubt on the relevance of existing destination image models, particularly when applied to individuals who already hold a less-than-positive perception. Hence, this study suggests four steps for enhancing cooperation between academics and practitioners: the use of a mixed team, re-examination of commercial (lagged) datasets, developing a combined set of hypotheses, and conducting rigorous analysis. The findings advance both theoretical understanding and practical strategy by showing that cognitive marketing messages may reinforce existing views but rarely overturn them. To support the market, academics should focus on conative destination image, develop segmentation tools to identify the target groups based on their overall destination image, and build dynamic destination image models that portray the differences between the groups and conditions. Full article

The application of intelligent video monitoring for natural resource protection and management has become increasingly common in recent years. To enhance safety monitoring during the grazing prohibition and rest period of grassland, this paper proposes a multi-object segmentation algorithm based on mid-infrared images for all-weather surveillance. The approach integrates the Image Timing Features–Gaussian Mixture Model (ITF-GMM) and Convolutional-UNet (Con-UNet) to improve the accuracy of target detection. First, a robust background modelling, i.e., the ITF-GMM, is proposed. Unlike the basic Gaussian Mixture Model (GMM), the proposed model dynamically adjusts the learning rate according to the content difference between adjacent frames and optimizes the number of Gaussian distributions through time series histogram analysis of pixels. Second, a segmentation framework based on Con-UNet is developed to improve the feature extraction ability of UNet. In this model, the maximum pooling layer is replaced with a convolutional layer, addressing the challenge of limited training data and improving the network’s ability to preserve spatial features. Finally, an integrated computation strategy is designed to combine the outputs of ITF-GMM and Con-UNet at the pixel level, and morphological operations are performed to refine the segmentation results and suppress noises, ensuring clearer object boundaries. The experimental results show the effectiveness of proposed approach, achieving a precision of 96.92%, an accuracy of 99.87%, an intersection over union (IOU) of 94.81%, and a recall of 97.75%. Furthermore, the proposed algorithm meets real-time processing requirements, confirming its capability to enhance small-target detection in complex outdoor environments and supporting the automation of grassland monitoring and enforcement. Full article

Background/Objectives: Ventricular tachycardia (VT) resulting in sudden cardiac death is common following a myocardial infarction (MI). Our objective was to evaluate the effects of an intracoronary (IC) administration of cardiosphere-derived cell secretome (S-CDCs) on VT inducibility and cardiac function in a swine model of MI. Methods: Fourteen pigs underwent endovascular MI model creation. At 4 weeks, saline (CON; 5 mL; n = 7) or S-CDCs (S-CDCs; 9.16 mg protein in 5 mL saline; n = 7) was blindly administered via the IC route. VT inducibility and magnetic resonance imaging (MRI) studies were performed both pre- and 4 months post-IC therapy, calculating left ventricular ejection fraction (LVEF), infarct size as a percentage of left ventricle (% MI), and left ventricular indexed end-diastolic and end-systolic volumes (LVEDVi, LVESVi). Results: While VT was inducible in 100% of the animals before IC therapy, at 4 months, the inducibility rate was lower in the S-CDCs group compared to the CON group (57% versus 100%, p = 0.05). Likewise, in the S-CDCs group, % MI was significantly lower than in the CON group (12 ± 3% versus 16 ± 3%, p = 0.03). LVEF (S-CDCs: 35 ± 10% versus CON: 29 ± 10%, p = NS), LVEDVi and LVESVi (S-CDCs: 83 ± 18 mL/m² and 56 ± 20 mL/m² versus CON: 88 ± 29 mL/m² and 64 ± 20 mL/m², p = NS) did not change. Conclusions: IC therapy with S-CDCs appears to reduce the development of post-MI VT. Furthermore, it suggests a beneficial effect on infarct size, reducing % MI in this experimental swine model. Full article

Nearshore ship detection is an important task in marine monitoring, playing a significant role in navigation safety and controlling illegal smuggling. The continuous research and development of Synthetic Aperture Radar (SAR) technology is not only of great importance in military and maritime security fields but also has great potential in civilian fields, such as disaster emergency response, marine resource monitoring, and environmental protection. Due to the limited sample size of nearshore ship datasets, it is difficult to meet the demand for the large quantity of training data required by existing deep learning algorithms, which limits the recognition accuracy. At the same time, artificial environmental features such as buildings can cause significant interference to SAR imaging, making it more difficult to distinguish ships from the background. Ship target images are greatly affected by speckle noise, posing additional challenges to data-driven recognition methods. Therefore, we utilized a Concurrent Single-Image GAN (ConSinGAN) to generate high-quality synthetic samples for re-labeling and fused them with the dataset extracted from the SAR-Ship dataset for nearshore image extraction and dataset division. Experimental analysis showed that the ship recognition model trained with augmented images had an accuracy increase of 4.66%, a recall rate increase of 3.68%, and an average precision (AP) with Intersection over Union (IoU) at 0.5 increased by 3.24%. Subsequently, an enhanced YOLOv7 algorithm (YOLOv7 + ESE) incorporating channel-wise information fusion was developed based on the YOLOv7 architecture integrated with the Squeeze-and-Excitation (SE) channel attention mechanism. Through comparative experiments, the analytical results demonstrated that the proposed algorithm achieved performance improvements of 0.36% in precision, 0.52% in recall, and 0.65% in average precision (AP@0.5) compared to the baseline model. This optimized architecture enables accurate detection of nearshore ship targets in SAR imagery. Full article

Background/Objectives: Multiple sclerosis (MS) is an autoimmune disease that damages the myelin sheath of the central nervous system, which includes the brain and spinal cord. Although MS lesions in the brain are more frequently investigated, MS lesions in the cervical spinal cord (CSC) can be much more specific for the diagnosis of the disease. Furthermore, as lesion burden in the CSC is directly related to disease progression, the presence of lesions in the CSC may help to differentiate MS from other neurological diseases. Methods: In this study, two novel deep learning models based on fractal architectures are proposed for the automatic detection and segmentation of MS lesions in the CSC by improving the convolutional and connection structures used in the layers of the U-Net architecture. In our previous study, we introduced the FractalSpiNet architecture by incorporating fractal convolutional block structures into the U-Net framework to develop a deeper network for segmenting MS lesions in the CPC. In this study, to improve the detection of smaller structures and finer details in the images, an attention mechanism is integrated into the FractalSpiNet architecture, resulting in the Att-FractalSpiNet model. In addition, in the second hybrid model, a fractal convolutional block is incorporated into the skip connection structure of the U-Net architecture, resulting in the development of the Con-FractalU-Net model. Results: Experimental studies were conducted using U-Net, FractalSpiNet, Con-FractalU-Net, and Att-FractalSpiNet architectures to detect the CSC region and the MS lesions within its boundaries. In segmenting the CSC region, the proposed Con-FractalU-Net architecture achieved the highest Dice Similarity Coefficient (DSC) score of 98.89%. Similarly, in detecting MS lesions within the CSC region, the Con-FractalU-Net model again achieved the best performance with a DSC score of 91.48%. Conclusions: For segmentation of the CSC region and detection of MS lesions, the proposed fractal-based Con-FractalU-Net and Att-FractalSpiNet architectures achieved higher scores than the baseline U-Net architecture, particularly in segmenting small and complex structures. Full article