Search Results (3,867)

The application prospects of photorealistic 3D reconstruction are broad in smart cities, cultural heritage preservation, and related domains. However, existing methods face persistent challenges in balancing reconstruction accuracy, computational efficiency, and robustness, particularly in complex scenes characterized by reflective surfaces, vegetation, sparse viewpoints, or large-scale structures. In this study, an enhanced 3D Gaussian Splatting (3DGS) framework that integrates three key innovations is proposed: (i) a depth-aware regularization module that leverages metric depth priors from the pre-trained Depth-Anything V2 model, enabling geometrically informed optimization through a dynamically weighted hybrid loss; (ii) a gradient-driven adaptive densification mechanism that triggers Gaussian adjustments based on local gradient saliency, reducing redundant computation; and (iii) a neighborhood density-based floating artifact detection method that filters outliers using spatial distribution and opacity thresholds. Extensive evaluations are conducted across four diverse datasets—ranging from architectures, urban scenes, natural landscapes with water bodies, and long-range linear infrastructures. Our method achieves state-of-the-art performance in both reconstruction quality and efficiency, attaining a PSNR of 34.15 dB and SSIM of 0.9382 on medium-sized scenes, with real-time rendering speeds exceeding 170 FPS at a resolution of 1600 × 900. It demonstrates superior generalization on challenging materials such as water and foliage, while exhibiting reduced overfitting compared to baseline approaches. Ablation studies confirm the critical contributions of depth regularization and gradient-sensitive adaptation, with the latter improving training efficiency by 38% over depth supervision alone. Furthermore, we analyze the impact of input resolution and depth model selection, revealing non-trivial trade-offs between quantitative metrics and visual fidelity. While aggressive downsampling inflates PSNR and SSIM, it leads to loss of high-frequency detail; we identify 1/4–1/2 resolution scaling as an optimal balance for practical deployment. Among depth models, Vitb achieves the best reconstruction stability. Despite these advances, memory consumption remains a challenge in large-scale scenarios. Future work will focus on lightweight model design, efficient point cloud preprocessing, and dynamic memory management to enhance scalability for industrial applications. Full article

We present a variable-rate learned image compression (LIC) model that integrates Transformer-based quantization–reconstruction (QR) offset prediction, entropy-guided hyper-latent quantization, and perceptually informed multi-objective optimization. Unlike existing LIC frameworks that train separate networks for each bitrate, the proposed method achieves continuous rate adaptation within a single model by dynamically balancing rate, distortion and perceptual objectives. Channel-wise asymmetric quantization and a composite loss combining MSE and LPIPS further enhance reconstruction fidelity and subjective quality. Experiments on the Kodak, CLIC2020 and Tecnick datasets show gains of +1.15 dB PSNR, +0.065 MS-SSIM, and −0.32 LPIPS relative to the baselines variable-rate method, while improving bitrate-control accuracy by 62.5%. With approximately 15% computational overhead, the framework achieves competitive compression efficiency and enhanced perceptual quality, offering a practical solution for adaptive, high-quality image delivery. Full article

This study investigated the effects of transcranial direct current stimulation (tDCS) over the dorsolateral prefrontal cortex (dlPFC) at three intensities (sham, 1 mA, 2 mA) on postural control, isokinetic strength, and cognitive performance in women with fibromyalgia (FM) and healthy controls (HCs). Using a double-blind, sham-controlled, crossover design, 26 participants (13 FM, 13 HC) completed sessions in randomized order, performing tasks under single- and dual-task conditions. Cognitive accuracy improved in both groups following 1 mA and 2 mA stimulation, particularly during single-task scenarios in static balance tasks. Notably, 2 mA tDCS reduced dual-task cost (DTC) in cognitive performance for the FM group, indicating decreased cognitive–motor interference. However, postural and strength outcomes showed no consistent intensity-dependent changes, with only selected nonlinear centers of pressure metrics (e.g., Lyapunov exponent, DFA) indicating possible modulation in FM. Isokinetic strength measures remained largely unaffected by tDCS across all intensities. Overall, the findings suggest that dlPFC-tDCS may selectively enhance cognitive function and reduce cognitive–motor interference in FM, especially under low-demand or higher-intensity stimulation conditions, while offering limited benefits for physical strength and balance. Full article

Demand-responsive transit (DRT) is a flexible transportation service that adapts routes and schedules based on real-time passenger needs, offering greater convenience than traditional fixed-route systems. DRT systems are highly dynamic and complex. Customer requests are often rejected due to operational constraints. Therefore, it is essential to identify and rank the factors that determine request acceptance or rejection. This study develops a Decision Tree Model (DTM) for vehicle dispatching in DRT, using the Korea National University of Transportation (KNUT) Chungju Campus as the study area. Elecle bicycle origin–destination (OD) data were first used to simulate DRT operations, and the resulting outputs were employed to train the DTM to classify passenger requests as “assign” or “reject.” The model considers key factors such as vehicle capacity, access time, Estimated Time of Arrival (ETA), waiting time, detour factor, and egress time. Based on 5-fold cross-validation, the detour factor was identified as the most influential variable across all fleet configurations, with mean importance values of 0.582 ± 0.055, 0.550 ± 0.047, and 0.447 ± 0.073 for the 1-, 2-, and 3-vehicle scenarios, respectively. The model achieved accuracies of 0.73 ± 0.02, 0.82 ± 0.04, and 0.83 ± 0.07, indicating improved performance with increasing fleet size. Error analysis revealed conservative behavior for one vehicle, balanced performance for two, and liberal over-assignment for three vehicles. Full article

Background: Chest radiography (CXR) is the most frequently performed radiological exam worldwide, but reporting backlogs, caused by a shortage of radiologists, remain a critical challenge in emergency care. Artificial intelligence (AI) triage systems can help alleviate this challenge by differentiating normal from abnormal studies and prioritizing urgent cases for review. This study aimed to externally validate TRIA, a commercial AI-powered CXR triage algorithm (NeuralMed, São Paulo, Brazil). Methods: TRIA employs a two-stage deep learning approach, comprising an image segmentation module that isolates the thoracic region, followed by a classification model trained to recognize common cardiopulmonary pathologies. We trained the system on 275,399 CXRs from multiple public and private datasets. We performed external validation retrospectively on 1045 CXRs (568 normal and 477 abnormal) from a teaching university hospital that was not used for training. We established ground truth using a large language model (LLM) to extract findings from original radiologist reports. An independent radiologist review of a 300-report subset confirmed the reliability of this method, achieving an accuracy of 0.98 (95% CI 0.978–0.988). We compared four ensemble decision strategies for abnormality detection. Performance metrics included sensitivity, specificity, accuracy, and area under the receiver operating characteristic curve (AUROC) with 95% CI. Results: The general abnormality classifier achieved strong performance (AUROC 0.911). Individual pathology models for cardiomegaly, pneumothorax, and effusion showed excellent results (AUROC of 0.968, 0.955, and 0.935, respectively). The weighted ensemble demonstrated the best balance, with an accuracy of 0.854 (95% CI, 0.831–0.874), a sensitivity of 0.845 (0.810–0.875), a specificity of 0.861 (0.830–0.887), and an AUROC of 0.927 (0.911–0.940). Sensitivity-prioritized methods achieving sensitivity >0.92 produced lower specificity (<0.69). False negatives were mainly subtle or equivocal cases, although many were still flagged as abnormal by the general classifier. Conclusions: TRIA achieved robust and balanced accuracy in distinguishing normal from abnormal CXRs. Integrating this system into clinical workflows has the potential to reduce reporting delays, prioritize urgent cases, and improve patient safety. These findings support its clinical utility and warrant prospective multicenter validation. Full article

Accurate land cover mapping is essential for environmental monitoring and agricultural management. Sentinel-2 imagery, with high spatial resolution and open access, provides valuable opportunities for operational classification. Convolutional neural networks (CNNs) have demonstrated state-of-the-art results, yet their adoption is limited by high computational demands and limited methodological transparency. This study proposes a lightweight CNN for soil–vegetation classification, in Dolj County, Romania. The architecture integrates three convolutional blocks, global average pooling, and dropout, with fewer than 150,000 trainable parameters. A fully documented workflow was implemented, covering preprocessing, patch extraction, training, and evaluation, addressing reproducibility challenges common in deep leaning studies. Experiments on Sentinel-2 imagery achieved 91.2% overall accuracy and a Cohen’s kappa of 0.82. These results are competitive with larger CNNs while reducing computational requirements by over 90%. Comparative analyses showed improvements over an NDVI baseline and a favorable efficiency–accuracy balance relative to heavier CNNs reported in the literature. A complementary ablation analysis confirmed that the adopted three-block architecture provides the optimal trade-off between accuracy and efficiency, empirically validating the robustness of the proposed design. These findings highlight the potential of lightweight, transparent deep learning for scalable and reproducible land cover monitoring, with prospects for extension to multi-class mapping, multi-temporal analysis, and fusion with complementary data such as SAR. This work provides a methodological basis for operational applications in resource-constrained environments. Full article

The occurrence of injuries due to unintentional falls becomes particularly dangerous in people with neurological disorders such as Parkinson’s disease (PD). The present study aimed to evaluate the influence of the Rock Steady Boxing (RSB) program’s single training units on body stability changes in elderly people with PD. Data from 18 patients (PG) and a similar-age 15-person control group without PD (CG) were used to analyze the collected study material. Postural stabilography was used to assess balance ability in two 30-second tests of standing on both feet with eyes open (EO) and closed (EC). The EO_CoP path length was significantly (p = 0.007) longer in the PG (266 ± 105 mm) compared to the CG (172 ± 32 mm), and similar differences were obtained for other parameters. PG measurements were taken over four consecutive weeks of RSB, both before and after each 90-minute training session. The lack of relevant differences be-tween measurements before vs. after for the PG may indicate the appropriate adaptation of exercisers to the applied loads. The probability of the compared parameters ranged from p = 0.586 to p = 0.999. Participation in RSB-based physical activity resulted in a deterioration in balance ability in the majority of participants immediately after exer-cise, but the results were characterized by a large spread, and the changes varied indi-vidually. Therefore, sports activities for PD must be adapted accordingly, taking into account the stage of the disease or the duration of the disease. Full article

Background and Objectives: Stratifying thyroid nodules according to malignancy risk is a crucial step in early diagnosis and patient care. Recently, deep learning techniques have emerged as powerful tools for medical diagnostics, particularly with convolutional neural networks (CNNs) applied to medical image classification. This study aimed to develop a new hybrid CNN model for classifying thyroid nodules using the TN5000 ultrasound image dataset. Materials and Methods: The TN5000 dataset includes 5000 ultrasound images, with 3572 malignant and 1428 benign nodules. To address the issue of class imbalance, the researchers applied an R-based anomaly data augmentation method and a GAN-based technique (G-RAN) to generate synthetic benign images, resulting in a balanced dataset for training. The model architecture was built on a pre-trained EfficientNet-B3 backbone, further enhanced with squeeze-and-excitation (SE) blocks and residual refinement modules to improve feature extraction. The task was to classify malignant nodules (labeled 1) and benign nodules (labeled 0). Results: The proposed hybrid CNN achieved strong performance, with an accuracy of 89.73%, sensitivity of 90.01%, precision of 88.23%, and an F1-score of 88.85%. The total training time was 42 min. Conclusions: The findings demonstrate that the proposed hybrid CNN model is a promising tool for thyroid nodule classification on ultrasound images. Its high diagnostic accuracy suggests that it could serve as a reliable decision-support system for clinicians, improving consistency in diagnosis and reducing human error. Future work will focus on clinical validation, explainability of the model’s decision-making process, and strategies for integration into routine hospital workflows. Full article

Purpose: This study aimed to develop a precise predictive model to assess the risk of radiation pneumonitis (RP) and three-year survival in patients with non-small cell lung cancer (NSCLC) following volumetric modulated arc therapy (VMAT). Radiomics features, ensemble stacking, and explainable artificial intelligence (XAI) were integrated to enhance predictive performance and clinical interpretability. Materials and Methods: A retrospective cohort of 221 NSCLC patients treated with VMAT at Kaohsiung Veterans General Hospital between 2013 and 2023 was analyzed, including 168 patients for RP prediction (47 with ≥grade 2 RP) and 118 patients for survival prediction (34 deaths). Clinical variables, dose–volume histogram (DVH) parameters, and radiomic features (original, Laplacian of Gaussian [LoG], and wavelet filtered) were extracted. ANOVA was used for initial feature reduction, followed by LASSO and Boruta-SHAP for feature selection, which formed 10 feature subsets. The data were divided at an 8:2 ratio into training and testing sets, with SMOTE balancing and 10-fold cross-validation for parameter optimization. Six models—logistic regression (LR), random forest (RF), support vector machine (SVM), k-nearest neighbors (KNN), XGBoost, and Ensemble Stacking—were evaluated in terms of the AUC, accuracy (ACC), negative predictive value (NPV), precision, and F1 score. SHAP analysis was applied to interpret feature contributions. Results: For RP prediction, the LASSO-selected radiomic subset (FR) combined with Ensemble Stacking achieved optimal performance (AUC 0.91, ACC 0.89), with SHAP identifying V40 Firstorder_Min as the most influential feature. For survival prediction, the FR subset yielded an AUC of 0.97, an ACC of 0.92, and an NPV of 1.00, with V10 Wavelet Firstorder_Min as the top contributor. The multimodal subset (FC+R) also performed strongly, achieving an AUC of 0.91 for RP and 0.96 for survival. Conclusions: This study demonstrated the superior performance of radiomics combined with Ensemble Stacking and XAI for the prediction of RP and survival following VMAT in patients with NSCLC. SHAP-based interpretation enhances transparency and clinical trust, offering a robust foundation for personalized radiotherapy and precision medicine. Full article

The rapid growth of technology has introduced robots into daily life, necessitating navigation frameworks that enable safe, human-friendly movement while accounting for social aspects. Such methods must also scale to situations with multiple humans and robots moving simultaneously. Recent advances in Deep Reinforcement Learning (DRL) have enabled policies that incorporate these norms into navigation. This work presents a socially aware navigation framework for mobile robots operating in environments shared with humans and other robots. The approach, based on single-agent DRL, models all interaction types between the ego robot, humans, and other robots. Training uses a reward function balancing task completion, collision avoidance, and maintaining comfortable distances from humans. An attention mechanism enables the framework to extract knowledge about the relative importance of surrounding agents, guiding safer and more efficient navigation. Our approach is tested in both dynamic and static obstacle environments. To improve training efficiency and promote socially appropriate behaviors, Imitation Learning is employed. Comparative evaluations with state-of-the-art methods highlight the advantages of our approach, especially in enhancing safety by reducing collisions and preserving comfort distances. Results confirm the effectiveness of our learned policy and its ability to extract socially relevant knowledge in human–robot environments where social compliance is essential for deployment. Full article

Background: Peripheral neuropathy (PN) refers to a spectrum of symptoms resulting from dysfunctions of the peripheral sensory, motor, and autonomic neurons. PN is associated with significant balance impairments and an increased risk of falls, contributing to reduced functional independence and quality of life. Although diabetic PN has been extensively investigated, there remains a lack of synthesized evidence regarding rehabilitation approaches for individuals with non-diabetic PN. This systematic review aims to evaluate the effectiveness of stabilometric platforms incorporating virtual reality (VR) and feedback (FB) in improving balance and related outcomes in patients with PN of various etiologies. Methods: This systematic review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines and the protocol registered in PROSPERO (CRD420251086625). Seven major databases (PubMed, Scopus, ScienceDirect, Cochrane, Web of Science, Springer, and Wiley) were searched from inception to April 2025. Studies including adult patients with non-diabetic PN undergoing balance rehabilitation using stabilometric platforms with VR and FB were considered. The methodological quality of the included studies was assessed using the PEDro scale, RoB2, and ROBINS-I V2 tools. Results: A total of six studies met the inclusion criteria, encompassing 133 participants with non-diabetic PN. Interventions involving specialized balance training platforms incorporating VR and FB demonstrated significant improvements in both static and dynamic balance and postural control, as well as a reduction in the risk of falling. These systems also showed favorable adherence rates to rehabilitation programs. However, variability in intervention protocols and outcome measures limited the ability to perform direct comparisons across studies. Conclusions: The use of stabilometric platforms appears to be a promising approach for balance rehabilitation in patients with non-diabetic PN. Despite the limited number of included studies, the results support their integration into rehabilitation programs for this patient population. Further large-scale, high-quality studies are needed to establish standardized protocols and confirm long-term efficacy. Full article

Although computer vision methods have advanced in construction helmet detection in recent years, their performance heavily depends on large-scale, class-balanced, and diverse annotated datasets. To address the high cost and labor-intensive nature of traditional data collection and annotation, this study introduces a novel helmet detection dataset named AIGC-HWD (Artificial Intelligence-Generated Content–Helmet Wearing Detection), automatically generated using generative AI tools. The dataset contains five categories of labels, supporting both helmet-wearing detection and color classification tasks. We evaluate the standalone performance of AIGC-HWD, as well as its augmentation effect when combined with the real-world dataset GDUT-HWD, using multiple algorithms, including YOLO v8, YOLO v10, YOLO 11, YOLO v11-MobileNet v4, YOLO v13, Faster R-CNN, and RT-DETR. Experimental results show that models trained solely on AIGC-generated images can achieve mAP@50 scores exceeding 0.7 and 0.8 on real-world images in two separate tests, demonstrating a certain level of generalization. When used for data augmentation alongside real-world images, the performance improves to varying degrees—by approximately 1% on the YOLO series, and by over 10% on the two-stage algorithm Faster R-CNN. These findings validate the potential of generative AI images for safety monitoring in construction scenarios and provide new insights into the integration of synthetic and real-world data. Full article

Background: Professional football requires more attention in planning work regimens that balance players’ sports performance optimization and reduce their injury probability. Machine learning applied to sports science has focused on predicting these events and identifying their risk factors. Our study aims to (i) analyze the differences between injury incidence during training and matches and (ii) build and classify different predictive models of risk based on players’ internal and external loads across four sports seasons. Methods: This investigation involved 96 male football players (26.2 ± 4.2 years; 181.1 ± 6.1 cm; 74.5 ± 7.1 kg) representing a single professional football club across four analyzed seasons. The research was designed according to three methodological sets of assessments: (i) average season performance, (ii) two weeks’ performance before the event, and (iii) four weeks’ performance before the event. We applied machine learning classification methods to build and classify different predictive injury risk models for each dataset. The dependent variable is categorical, representing the occurrence of a time-loss muscle injury (N = 97). The independent variables include players’ information and external (GPS-derived) and internal (RPE) workload variables. Results: The Kstar classifier with the four-week window dataset achieved the best predictive performance, presenting an Area Under the Precision–Recall Curve (AUC-PR) of 83% and a balanced accuracy of 72%. Conclusions: In practical terms, this methodology provides technical staff with more reliable data to inform modifications to playing and training regimens. Future research should focus on understanding the technical staff’s qualitative vision of predictive models’ in-field applicability. Full article

Long-tailed image classification remains challenging for vision–language models. Head classes dominate training while tail classes are underrepresented and noisy, and short prompts with weak text supervision further amplify head bias. This paper presents TASA, an end-to-end framework that stabilizes textual supervision and enhances cross-modal fusion. A Semantic Distribution Modulation (SDM) module constructs class-specific text prototypes by cosine-weighted fusion of multiple LLM-generated descriptions with a canonical template, providing stable and diverse semantic anchors without training text parameters. Dual-Space Cross-Modal Fusion (DCF) module incorporates selective-scan state–space blocks into both image and text branches, enabling bidirectional conditioning and efficient feature fusion through a lightweight multilayer perceptron. Together with a margin-aware alignment loss, TASA aligns images with class prototypes for classification without requiring paired image–text data or per-class prompt tuning. Experiments on CIFAR-10/100-LT, ImageNet-LT, and Places-LT demonstrate consistent improvements across many-, medium-, and few-shot groups. Ablation studies confirm that DCF yields the largest single-module gain, while SDM and DCF combined provide the most robust and balanced performance. These results highlight the effectiveness of integrating text-driven prototypes with state–space fusion for long-tailed classification. Full article

Accurate prediction of bridge deterioration is essential for optimizing inspection schedules, prioritizing maintenance activities, and ensuring infrastructure safety. This study developed machine learning-based predictive models to estimate the deterioration states of bridge decks, superstructures, and substructures using a comprehensive dataset from the Colorado National Highway System spanning 2014 to 2024. Structural, operational, and environmental parameters including freeze–thaw cycles, precipitation, condensation risk, and extreme temperatures were incorporated to capture both design-driven and climate-driven deterioration mechanisms. Decision Tree, Random Forest, and Gradient Boosting classifiers were trained and evaluated using Balanced Accuracy, Matthews Correlation Coefficient, Cohen’s Kappa, and macro-averaged F1-scores, with class imbalance addressed via SMOTETomek resampling. The Gradient Boosting classifier achieved the highest predictive performance, with balanced accuracy exceeding 97% across all components. Feature importance analysis revealed that sufficiency rating, year of construction, and environmental stressors were among the most influential predictors. Incorporating environmental variables improved predictive accuracy by up to 4.5 percentage points, underscoring their critical role in deterioration modeling. These findings demonstrate that integrating environmental factors into machine learning frameworks enhances the reliability of deterioration forecasts and supports the development of climate-aware asset management strategies, enabling transportation agencies to proactively plan maintenance interventions and improve infrastructure resilience. Full article