Search Results (3,143)

A polyurethane coating represents the conventional solution for protecting and finishing visible wooden furniture surfaces. However, the increasing use of additive manufacturing has introduced visible 3D-printed polymer components whose surface mechanical performance is relevant to their application in furniture. This study investigated the effect of short-term exposure to 60 °C for 1 h on the surface mechanical properties of a pigmented polyurethane coating applied to oak wood and 3D-printed PLA, ABS-T and PET-G components. Specimens were evaluated under laboratory conditions (20 °C) and immediately after exposure to 60 °C, while their surfaces remained at an elevated temperature. Impact resistance, abrasion resistance, and scratch resistance using a tungsten carbide tip were determined according to the relevant standards, and the surface damage was assessed by visual inspection and digital microscopy. The polyurethane coating exhibited the smallest impact indentation diameter but showed earlier crack initiation and lower abrasion resistance than the 3D-printed polymers. Among the investigated polymers, ABS-T provided the most balanced combination of impact resistance, abrasion resistance and surface hardness, whereas PLA exhibited the greatest dimensional changes after exposure to 60 °C. Microscopic analysis revealed surface defects that were not detectable by visual inspection, demonstrating the value of digital microscopy for detecting subtle surface damage. The results indicate that ABS-T is a promising material for visible furniture components exposed to short-term elevated temperatures, while PET-G should be used with caution in applications exposed to radiant heat or direct sunlight. Full article

Background/Objectives: Reported performance on public brain tumour MRI benchmarks is hard to interpret because of near-duplicate train/test overlap, ImageNet pretraining bias, and single-seed evaluation. We address this with a leakage-aware evaluation protocol and a compact model trained entirely from scratch. Methods: WICA-Net-M is a 2.47 M-parameter CNN whose gated Haar Discrete Wavelet Transform (DWT) separates low- and high-frequency components and fuses them through a learnable gate, complemented by Coordinate Attention. We evaluate it on the standard and image-level deduplicated splits of the Nickparvar brain tumour MRI dataset under a three-seed, leakage-aware protocol, benchmark it against five ImageNet-pretrained baselines and conduct a near-duplicate overlap audit against BRISC 2025. Results: WICA-Net-M reaches 99.42 ± 0.16% accuracy on the standard V1 split and 95.25 ± 0.32% accuracy/95.17 ± 0.31% macro F1 on the deduplicated V2 split, closely matching five ImageNet-pretrained baselines (95.17–95.67%) with fewer parameters, with sub-half-point differences across three seeds indicating comparable rather than superior accuracy. The audit identifies 861 exact SHA-256 pairs involving 857 of the 1000 BRISC test images against the full Nickparvar collection. Alongside perceptual-hash candidate pairs, this exact overlap shows that BRISC cannot serve as independent external validation. Conclusions: The descriptive 4.17-point V1-to-V2 difference reflects the combined effects of duplicate removal, class rebalancing, and altered sample composition, with none separable from public releases, though it shows that a scratch-trained compact model can approach pretrained performance on the controlled split. Patient-level leakage remains unresolved because patient identifiers are unavailable. Leakage-aware, multi-seed evaluation should be standard before clinical translation. Full article

This survey presents a structured review of deep learning-based techniques for image data hiding, proposing a three-paradigm taxonomy organized by the method’s operational relationship to the carrier image. We classify existing methods into modification-based, synthesis-based, and logic-based approaches. In the modification-based tier, we trace the architectural progression from foundational Convolutional Neural Networks and Generative Adversarial Networks to high-capacity Invertible Neural Networks and Transformers, analyzing their distinct trade-offs between embedding capacity, imperceptibility, and robustness. In the synthesis-based tier, we examine how Diffusion Probabilistic Models and generative adversarial frameworks reframe data hiding as a carrier generation problem rather than a pixel editing task. This paradigm encompasses both generative steganography (where carriers are synthesized from scratch) and proactive watermarking (where provenance is embedded during AI content generation). In the logic-based tier, we review zero-watermarking and coverless steganography, where ownership is established through feature extraction and semantic mapping without modifying any image, a critical property for sensitive domains such as medical imaging. Finally, we identify four persistent infrastructure gaps: benchmarking fragmentation, narrow robustness evaluation, domain generalization failures, and computational infeasibility that prevent real-world deployment despite architectural progress, and we propose concrete research directions. Full article

Background: Multicomponent exercise promotes healthy aging, but evidence in community-dwelling older women remains limited. This study examined pre–post changes in functional fitness and cardiovascular health following participation in a six-month multicomponent exercise program. Methods: Thirty-five community-dwelling older women (71.39 ± 6.82 years) participated in a quasi-experimental pre–post intervention study. Participants completed supervised multicomponent exercise sessions twice weekly for six months. Anthropometric, cardiovascular, and functional fitness outcomes were assessed before and after the program. Results: Reductions were observed in body mass (68.03 ± 11.39 vs. 66.29 ± 10.02 kg; p < 0.001) and body mass index (26.56 ± 3.71 vs. 25.90 ± 3.22 kg·m⁻²; p < 0.05). Lower post-test values were also observed for systolic blood pressure (125 ± 9 vs. 121 ± 10 mmHg; p = 0.026), resting heart rate (85 ± 12 vs. 77 ± 11 bpm; p = 0.008), and double product (10,359 ± 1899 vs. 9701 ± 1705 mmHg·bpm; p = 0.027). Handgrip strength was higher at post-test in both the dominant hand (20.98 ± 5.47 vs. 22.53 ± 6.31 kgf; p = 0.007) and non-dominant hand (19.23 ± 4.99 vs. 21.20 ± 5.61 kgf; p < 0.001). Functional aerobic capacity was also higher at post-test (95.74 ± 21.35 vs. 110.41 ± 22.86 steps; p < 0.001), while flexibility values were higher in the left Chair Sit-and-Reach Test (−15.92 ± 10.48 vs. −12.97 ± 10.91 cm; p = 0.001) and left Back Scratch Test (−3.65 ± 6.82 vs. 4.32 ± 7.83 cm; p < 0.001). Conclusions: Participation in a twice-weekly multicomponent exercise program for six months was associated with favorable pre–post changes in functional fitness and cardiovascular health in community-dwelling older women. Full article

The e-commerce sector offers unprecedented growth opportunities and reshapes traditional business models, which are enhanced by the increasing integration of gamification elements. This development emphasizes the urgency of comprehensively analyzing this topic and generating first insights into potential drivers of conversion rates and shopping cart abandonments. Although gamification is already widely used in practice, there is a lack of comparative scientific research. A mixed-methods approach is utilized, combining qualitative expert interviews (five experts from heterogeneous backgrounds) and discrete choice experiments with consumers (102 participants and five choices each across both experiments). A relative ranking of common gamification elements is empirically established both in the context of increasing conversion rates and reducing shopping cart abandonments. Both contexts lead to inherently different rankings, implying differences in the underlying mechanics. Furthermore, the results imply that gamification in the context of shopping cart abandonments is a much more complex situation than its deployment in other parts of the e-commerce platform. While games of chance, except scratch cards, are perceived negatively across both contexts, when it comes to shopping cart abandonments, passive elements, such as progress bars and timers, are distinctly preferred over active elements, such as mini games. The findings hint at the need for careful selection and adaptation of gamification strategies in accordance with the objectives of consumers during the different phases of the customer journey. Full article

Synthesizing realistic radio-frequency signals is important for data augmentation in wireless systems, yet labeled in-phase/quadrature (I/Q) measurements are expensive to collect, and diffusion models typically demand large per-class training sets. This paper studies whether a diffusion model trained on one modulation scheme can be transferred to unseen schemes with minimal data and a minimal parameter budget. We formulate the adaptation as a minimal-cardinality parameter-increment problem and adapt a conditional diffusion transformer, pretrained on QPSK signals from the RML2018.01a corpus, to three target modulations (OQPSK, 16PSK, and 32QAM) through low-rank adaptation (LoRA) of its attention projections under a composite objective that couples waveform reconstruction with spectral and autocorrelation constraints, updating only 0.496% of the 9.9-million-parameter backbone. With only 50 target signals, the adapted models attain a periodogram cosine similarity of 0.985 against 0.705 for from-scratch training and match from-scratch models given 1000 signals: a twenty-fold gain in sample efficiency. The transferred models also match or exceed reference target experts trained on up to 333 times more target data on every spectral and autocorrelation metric. Ablations on two target schemes show that sources with dense phase manifolds transfer best irrespective of family labels and that the generation quality is insensitive to the LoRA rank between 1 and 16, pointing to a low-dimensional adaptation subspace. Downstream classifier experiments confirm that the synthetic signals help few-shot modulation classification, and they also bound the claim: conventional label-preserving transforms remain stronger at very small sample sizes, downstream utility does not track spectral fidelity, and the generated signals under-represent the noise floor in the low-SNR regime. The main study uses RML2018.01a; a cross-corpus check on the independently generated Sig53 corpus reproduces the transfer advantage with compressed margins, and validation on over-the-air recordings remains future work. Full article

CrN coatings prepared by physical vapor deposition may fail prematurely in localized regions during machining due to stress and high temperature. To address this problem, three groups of CrN coatings with different bias voltages were deposited on cemented carbide tools using cathodic arc ion plating. The influence of coating microstructure on cutting performance was investigated. The surface and cross-sectional morphologies of the coatings were observed using scanning electron microscopy. The crystal structures of the coatings were characterized using X-ray diffraction. The nanohardness values and elastic moduli of the coatings were measured using nanoindentation, and the coating–substrate adhesion was evaluated using a scratch tester. The residual stresses of the coatings were determined using the substrate bending method. The cutting performance of the coated tool was tested using a Computer Numerical Control (CNC) machine tool. The cutting process using coated tools was simulated using ABAQUS 6.14 software. The results indicated that, as the bias voltage increased, the droplets on the surface of the CrN coating became smaller and the structure became denser, but the thickness gradually decreased. The hardness increased from 18.59 GPa at 100 V to 24.57 GPa at 200 V, the bonding force increased from 72.6 N to 78.5 N, and the residual stress also increased to −2.22 GPa. Cutting tests showed that the cutting performance of the 200V-CrN coating was nearly twice that of the 100V-CrN coating. According to the microstructure and finite element simulation results after cutting, the 200V-CrN coating with a dense structure and higher residual compressive stress exhibited better stress-absorption capacity and was less susceptible to wear or fracture during the cutting process, thereby extending tool life. Full article

Accurate defect instance segmentation is essential for automated quality inspection of piezoelectric crystals, where scratches, stains, and other surface defects often show large scale variation, irregular morphology, and weak contrast against complex backgrounds. Existing CNN-based segmentation models capture local texture effectively but are limited in global context modeling, whereas Transformer-based models often require stronger local detail preservation for fine industrial defects. To address these limitations, this paper proposes a hybrid bidirectional bridging CNN–Transformer network, named HBCTNet, for piezoelectric crystal defect instance segmentation. HBCTNet introduces a Convolution-Transformer Bidirectional Bridging Block (CTB) that enables lightweight two-way interaction between convolutional local features and Transformer global representations. A Frequency and Spatial Convolution (FSC) module is designed to enhance defect-related details in both spatial and frequency domains, and a Multi-Scale Feature Pyramid Network (MS-FPN) fuses features across scales. On the reported image partition of the self-constructed dataset, HBCTNet-S obtains 92.3% box AP and 89.9% mask AP with 12.67 M parameters and 48.1 G FLOPs. Its mask AP is 2.1 percentage points higher than YOLO11-SEG and 2.4 percentage points higher than YOLOv9-SEG under the reported comparison. These results support the effectiveness of the architecture within the evaluated dataset; inference speed and cross-domain generalization remain to be assessed. Full article

To enable rapid, non-destructive identification of damage to Yali pears, this study proposes a detection method that integrates short-wave infrared hyperspectral imaging (1000–2500 nm), regularised linear discriminant analysis (R-LDA) and a lightweight convolutional neural network (CNN). The experiments utilised 180 Yali pears (60 each of healthy, with Mechanical scratch and Compression damage specimens) as training samples, whilst a further 300 independent fruits (150 healthy and 150 damaged) were used for fruit-level sorting validation. Following pre-processing using Principal Component Analysis (PCA) to eliminate multicollinearity, the classification performance of three feature extraction strategies—PCA, Independent Component Analysis (ICA) and R-LDA—was compared when combined with the same lightweight CNN. The results indicate that R-LDA’s Fisher discrimination criterion (5.0275) and separation index (2.3543) were both superior to those of PCA and ICA, and its dimension-reduced features exhibited stronger inter-class separability. In pixel-level testing, the R-LDA + lightweight CNN achieved recognition accuracies of 98.60 per cent and 99.32 per cent for Compression damage and background, respectively, and 93.54 per cent and 96.33 per cent for Mechanical scratch and intact tissue, respectively. In a validation study involving the sorting of 300 independent fruits, this method achieved a recall rate of 100.00% for damaged fruits (zero false negatives), with precision and F1 scores of 97.00% and 97.09% respectively, both of which outperformed PCA combined with a lightweight CNN and ICA combined with a lightweight CNN. The above results indicate that the combination of R-LDA discriminant dimensionality reduction and a lightweight CNN can effectively reduce redundancy in hyperspectral data whilst maintaining a high damage detection rate, thereby providing a viable solution for the rapid, non-destructive detection of post-harvest damage in Yali pears. Full article

Foundation models for Earth observation are commonly transferred to downstream tasks without explicit attention to the preprocessing mismatch between pretraining and target distributions. This study isolates preprocessing mismatch as a controlled experimental factor in transferring the Hydro multispectral foundation model, a Swin Transformer V2 Base encoder pretrained with SimMIM on Sentinel-2 Level-2A water-body imagery, to the Marine Debris and Oil Spill (MADOS) marine pollution benchmark dataset, processed through ACOLITE Rayleigh reflectance and providing 11 of the 12 spectral bands used during pretraining. The two datasets are therefore produced by different atmospheric correction algorithms under different reflectance conventions, and the resulting per-band statistical discrepancy is quantified as the starting point of the analysis. Three preprocessing dimensions are then systematically varied while all other settings are held constant: input normalisation, spectral band adaptation for the missing B09, and encoder transfer mode. From this, four findings emerge. Normalisation mismatch between training and inference is the single largest source of performance degradation, reducing the mean Intersection over Union (mIoU) by 0.458, more than seven times the largest radiometric perturbation tested. A zero-parameter Frobenius-matched column crop of the patch embedding adapts the 12-band pretrained encoder to the 11-band target, at least as effectively as any learnt linear or nonlinear adapter, at a lower cross-seed variance. Under limited target supervision (1433 training patches against an 87.9 million-parameter encoder), freezing the encoder outperforms both fine-tuning in full and random initialisation training from scratch. The gains of partial unfreezing are attributable to augmented training (very simple copy–paste (VSCP) augmentation, exponential moving average (EMA), and test-time augmentation (TTA)) rather than to encoder adaptation. With matched preprocessing, the frozen encoder reaches 0.600 mIoU and matches the published MariNeXt baseline within seed variability. Mechanistic analysis via band-occlusion attribution and feature-space separability shows that input normalisation determines which spectral bands the encoder relies upon, with the magnitude of the shift correlated to the per-band gap between the source and target distributions. Operationally, preprocessing alignment, rather than architectural modification, carries most of the practical effort in transferring a multispectral foundation model to marine surface segmentation. These results are established for a single encoder–benchmark pair under limited target supervision. The mechanism they identify is more portable than the magnitude reported. A frozen encoder’s representations remain bound to the normalisation statistics of its pretraining dataset, so any transfer that departs from these statistics at inference is predicted to degrade sharply in proportion to the per-band distance between the two distributions. Full article

Vision-based traffic accident anticipation is critical for active vehicle safety systems, yet existing architectures frequently conflate performance gains with heavy parameter scaling optimized from scratch on compact domains. Consequently, during real-time inference, these frameworks suffer from severe prediction volatility and early triggering biases that induce dangerous control instability. To address these limitations, this paper shifts the research focus away from network modifications toward a highly controlled, strategy-driven training pipeline executed under a completely invariant spatial–temporal neural backbone. Our proposed paradigm establishes a robust framework through three decoupled milestones. First, an out-of-domain initialization strategy transferred generalized driving kinetics from a large-scale sequence domain (Mapillary) to serve as a stable temporal anchor. Second, a target-domain generative enrichment step injected synthetic nighttime scenes to decouple hazard features from low-light ambient noise. Third, progressive temporal supervision paradigm scaling targeted labels monotonically to align with continuous kinetic risk accumulation. Overall evaluations on the Car Crash Dataset (CCD) benchmark demonstrate that the fully integrated configuration (C4) pipeline achieves 69.89% in frame-level Mean Average Precision (mAP), which is an improvement of +22.81 percentage points over the baseline configuration. Continuous temporal measurements prove that our framework can adapt to tracking volatility, compressing Temporal Confidence Variance to 0.00328, and dropping the Prediction Instability Count to 0.66. While hyper-sensitive baselines report early raw latency averages driven by premature trigger noise, our model purposefully filters this early-frame variability to deliver a secure warning profile, achieving an absolute zero false alarm rate (FAR = 0.00%) across evaluated non-hazardous driving sequences, establishing the sequence-level trustworthiness required for practical autonomous deployment. Full article

Time-series foundation models are designed to continue observed trajectories, whereas model predictive control (MPC) must compare counterfactual rollouts under candidate future actions. A forecaster whose predictions are invariant to those actions cannot rank controls by predicted plant response, irrespective of forecast error. We formalize this forecasting-to-control gap and convert pretrained Chronos-2 into an action-conditioned dynamics model. Chronos-2 Hybrid combines low-rank target-domain adaptation, separate process-history and future-decision pathways, and gated residual fusion. Removing future-decision conditioning degraded action-response fidelity in all ten Quadrotor seeds. In bidirectional transfer between two 660 MW thermal power units, the unit-balanced area under the sulfur dioxide trajectory root-mean-squared error (RMSE) learning curve was 14.95% lower than that of an input-matched scratch model; at 6 and 24 h of target labels, the RMSEs were 29.0% and 18.5% lower. A separate Benchmark Simulation Model No. 1 study showed the same low-budget pattern under long process memory. Across eight outcome independently matched desulfurization system episodes, the normalized emission–effort performance index was 29.8–69.6% lower during strategy-assisted operation. These results identify the action-conditioned interface required to use a pretrained forecaster in MPC and show why forecast error alone is insufficient to assess control readiness. Full article

This qualitative case study explores how two elementary classrooms of predominantly Pacific Islander students (the majority of whom identified as Native Hawaiian) worked toward developing their cultural competence through participation in an interdisciplinary classroom-based learning experience focused on Hawaiian culture-based STEAM and maker learning. Over a four-unit series focused on wildfire prevention, watersheds, and traditional mo‘olelo (storytelling), students engaged in collaborative, place-based learning. Findings reveal that cultural and academic competence were fostered through three primary mechanisms: integrating ancestral storytelling as a pedagogical tool; grounding academic content in hyper-localized ecological contexts; and navigating the “productive struggle” caused by the lack of any Pacific Island cultural representation in standard digital tools like Scratch and Canva. By adapting technologies to reflect their own identities, students simultaneously advanced their digital literacies and academic skills in science and language arts while honouring their heritage. This study underscores the importance of viewing cultural and STEAM learning as concurrent, mutually reinforcing processes. Our bounded case in particular highlights the possibility of how, when educators centre culture-based learning, classrooms can become spaces of cultural reclamation while supporting necessary academic competencies, helping to bridge the gap between student identity and the compulsory educational system. Full article

Metal corrosion shortens the service life of engineering materials and can cause structural failure and serious safety hazards. Conventional polymer coatings rely mainly on passive barrier protection and often provide limited protection after mechanical damage. Here, we developed a smart epoxy coating containing flake-like CeO₂–melamine (CeO₂-MA) composite fillers and systematically evaluated the effect of filler content on corrosion protection. Among the tested formulations, the 0.5 wt.% CeO₂-MA/epoxy coating exhibited the best overall corrosion protection. Following 72 h of immersion under artificial-defect conditions, its low-frequency impedance modulus (|Z|_0.01Hz) was approximately 3.3 times that of the epoxy control. This improved protection may be associated with the complementary effects of melamine adsorption and the formation of cerium-containing deposits in damaged regions. After thermal treatment at 60 °C for 10 min, the coating exhibited pronounced scratch narrowing with only minor residual grooves, along with partial recovery of electrochemical barrier performance. These findings support a multifunctional coating strategy, although direct validation of the inhibition mechanism and longer-term evaluation of the recovered barrier performance remains necessary. Full article

Background: In our companion study, the gingival mesenchymal stem cell (GMSC) secretome suppresses oxidative stress and induces apoptosis in primary oral squamous cell carcinoma (OSCC) cells, where those wet-lab results are themselves reported as preliminary; here we ask whether it also engages a proliferation × migration programme in patient tissue. The two arms differ in read-out type (apoptosis and redox there, transcript abundance here), not in opposed function. Methods: Primary OSCC cells received GMSC-conditioned medium (GMSC-CM) or indirect Transwell co-culture, assayed by RT-qPCR (VEGFA, TGFB1, MMP9, CXCL12, CCND1, PCNA, MYC, EGFR), MTT, and scratch-wound migration. Thirteen computational layers plus a GeoMx spatial verify-and-decide layer were applied to public data: TCGA-HNSC, GEO, CPTAC, single-cell inference, prognostic modelling, DepMap and LINCS L1000. Results: All eight transcripts were raised in all three arms (16 of 24 comparisons significant), indicating a pro-tumour transcriptional shift; metabolic activity was unchanged and wound closure was reduced under conditioned medium (p < 0.01). Of 1358 genes significant in all three modalities, 1219 (89.8%, 95% CI 88.0–91.3%) share direction against 25.0% expected by chance (3.59-fold; exact binomial p below machine precision). The pre-registered oral-cavity signature did not validate externally (GSE41613; C = 0.562), and no ligand–receptor pair survived permutation calibration. Conclusions: GMSC paracrine exposure induces a pro-tumour transcriptional programme in primary OSCC cells that replicates at patient level, without demonstrated functional or therapeutic consequence; gefitinib is nominated only as a hypothesis-generating candidate. The evidence rests on three primary cultures (n = 3), one GMSC donor preparation, one 24 h time point, a single reference gene, no cell-line authentication and no test of gefitinib; the study programme has concluded, and these experiments cannot be performed now. Full article