Search Results (912)

Human pose estimation (HPE) has advanced rapidly with deep learning, enabling a transition from specialized sensing and multi-view systems toward monocular RGB-based approaches. These developments have expanded applications in healthcare, robotics, sports analytics, and human–computer interaction. However, the growing diversity of deep learning paradigms, ranging from convolutional and recurrent models to graph-based and Transformer-based approaches, has resulted in a fragmented literature, making it difficult to systematically compare methods and guide system design. This paper addresses this challenge by providing a comprehensive survey of deep learning-based monocular HPE methods published over the past decade and introducing a unified modular framework. The proposed framework organizes HPE systems into six modular estimation paradigms, including single-image-based estimation, multi-frame-based estimation, Top-Down and Bottom-Up pose estimation strategies, 2D-to-3D pose reconstruction, and direct 3D estimation. Each module is analyzed in terms of representative approaches, design trade-offs, and practical considerations, supported by algorithmic formulations that outline the computational pipeline at each stage. Unlike prior surveys that primarily catalog methods or report benchmark results in isolation, this work emphasizes how component-level design choices relate to overall system performance. The paper summarizes performance trends on benchmarks including Human3.6M, COCO, and MPII, highlighting persistent challenges such as occlusion and viewpoint variation, and outlines future research directions including interaction-aware modeling, efficient deployment, and improved robustness under real-world conditions. Full article

Zero-shot remote sensing is attractive for scene classification because new regions, sensors, and label taxonomies often appear before sufficient annotated data are available for supervised adaptation. We present OATS-RS, an inference-centric framework that keeps a remote sensing vision–language model (VLM) backbone frozen and improves zero-shot decisions through ontology-aware prompt construction, hierarchical and contrastive scoring, adaptive multi-view aggregation, unlabeled transductive refinement, ambiguity-aware local re-ranking, and selective prediction. The method targets the common remote sensing regime in which neighboring classes such as annual crop, permanent crop, forest, pasture, herbaceous vegetation, river, and sea or lake overlap strongly in red–green–blue (RGB) appearance, meaning that they require more than a single class-name prompt. On the supplied final EuroSAT RGB evaluation with a GeoRSCLIP Contrastive Language–Image Pre-training (CLIP)-family Vision Transformer Base with 32 × 32-pixel patches (ViT-B-32) backbone, the complete pipeline obtains top-1 accuracy of 0.522, balanced accuracy of 0.522, macro-averaged F1 score (macro-F1) of 0.535, and top-3 accuracy of 0.887. The strongest classes are industrial area, residential area, river, highway, and pasture, whereas the weakest classes remain herbaceous vegetation and several fine-grained vegetation categories. Selective prediction increases accepted-example accuracy to 0.538 at 0.934 coverage, but the expected calibration error (ECE) remains high at 0.384. These results support a qualified conclusion: ontology-guided zero-shot inference can already recover useful semantic shortlists for structured remote-sensing scenes, but fine-grained natural-class disambiguation, calibrated confidence, multi-dataset transfer, component-level ablations, and measured runtime remain essential before dependable deployment claims can be made. Full article

Biomimetic design has been used to reduce the high operating resistance of agricultural soil-engaging components, thereby lowering energy consumption. However, most existing contour-based structural biomimetic designs rely on a single or a few biological samples, making the resulting designs susceptible to individual variation and randomness in sample selection. To address this issue, this study used the abdomen of antlion larvae as a biological prototype. Abdominal contours of 85 antlion larvae were extracted from the front, top, and side views, and elliptic Fourier descriptors (EFDs) were used for contour normalization, averaging, and reconstruction to obtain population mean contours. Seven biomimetic wedge specimens were designed based on the population mean contours, and vertical penetration and horizontal cutting tests were conducted in two different media. The results showed that in the vertical penetration tests, the B-FT specimen, which integrated contour features from the front and top views, exhibited the best drag-reduction performance. Its average penetration resistance decreased by 44.26% and 32.81% in quartz sand and loam soil, respectively. In the horizontal cutting tests, the B-FTS specimen, which integrated contour features from all three views, showed the lowest average cutting resistance, with reductions of 17.62% and 36.47%, respectively. The FTS contour features were further applied to the biomimetic design of a subsoiler tine and validated by discrete element method (DEM) simulation and soil bin tests. Compared with the standard subsoiler tine, the biomimetic subsoiler tine reduced draft force by 11.57% in the simulation and by 12.61% in the soil bin test. These results demonstrate the drag-reduction effectiveness of population mean contours and provide a statistically grounded geometric reference for the biomimetic low-resistance design of agricultural soil-engaging components. Full article

Background/Objectives: This study aimed to comparatively analyze online information regarding dental implants and dentures utilizing text-mining techniques. Methods: An automated text-mining program was employed to collect and process data using the Korean keywords for “implant” and “denture.” Data sources included major search engines, social networking services, and YouTube (Google LLC, Mountain View, CA, USA). A total of 9941 data points for dental implants and 9783 for dentures were retrieved. The analytical approach included word cloud generation, term frequency-inverse document frequency (TF-IDF) analysis, semantic network analysis, and sentiment analysis. Results: For implants, “dental clinic,” “treatment,” “surgery,” and “insurance” emerged as highly relevant keywords. In contrast, queries regarding dentures frequently included the term “implant,” alongside top-ranking, age-related terms such as “abnormality” and “discomfort.” TF-IDF analysis revealed that “surgery” and “procedure” ranked higher for implants, whereas “insurance” ranked higher for dentures. Sentiment analysis indicated a predominantly positive public perception of implants (63.09% positive, 36.91% negative), whereas dentures elicited a largely negative sentiment (40.70% positive, 59.30% negative). Conclusions: The text-mining analysis revealed distinct public perceptions regarding the two treatments. Implants were primarily associated with surgical procedures and positive sentiments, whereas dentures were more closely linked to insurance considerations and negative experiences. Full article

The field of pose estimation is a major problem in computer vision, enabling the direct transformation of an input image into a hierarchical representation of the human skeleton for application in the fields of virtual/augmented reality and human–machine interaction tasks. Research in this field has exploded between 2018 and 2025, with traditional taxonomies such as 2D versus 3D or top-down versus bottom-up no longer sufficient to capture the essence of the evolution of ideas. To solve this problem, we propose a conceptual review in the field of pose estimation, focusing on the intellectual evolution of methods and architecture rather than the standard flat classifications of papers. We divide recent advances into five structural pillars: Representation, which traces the evolution from pixel coordinate regression to heatmaps and probabilistic representation; Architecture, which analyzes the transition from multi-stage CNNs to transformers and state space models (SSMs); Ambiguity and Generalization, which analyzes how self-supervised, uncertainty-aware, and diffusion models address 3D depth ambiguity, occlusion, and domain gaps by modeling multiple plausible poses and reducing dependence on fully supervised in-the-wild 3D labels; Context Extension, which covers temporal dynamics, multi-view fusion, and potential sensors; and Applications, which links algorithms to efficiency, privacy, and foundation models. By providing an in-depth detailing of these pillars, we provide a unified view of the evolution of research paradigms that define human pose estimation and enable the identification of future problems and solutions in pose estimation and human-centered tasks. Full article

Transitioning toward efficient and digital industrial design requires preliminary tools that support early decision-making in plant layout studies. This qualitative and exploratory study analyzes an Artificial Intelligence (AI)-assisted and Augmented Reality (AR)-supported workflow within the Systematic Layout Planning (SLP) framework for the preliminary spatial evaluation of a cosmetic emulsion production plant. The study was developed as a case study based on a previously reported layout for obtaining cosmetic emulsions from Amazonian oils. A top-view layout was examined through structured prompts aligned with SLP criteria, including product journey, activity relationships, relational diagrams, and space requirements. ChatGPT was used only as a qualitative reasoning assistant, without optimization, prediction, mathematical modeling, or algorithmic functions. After the AI-assisted review, the refined layout was represented in three dimensions and visualized through AR in a real environment. The results identified potential improvements related to operational flow, traceability, critical area relationships, and spatial organization. AR-assisted visualization provided preliminary visual evidence of compatibility between the refined layout and the selected site, supporting an early review of circulation, access, and volumetric behavior. The sequential integration of SLP, AI, and AR is proposed as an exploratory workflow for early-stage layout evaluation, pending future quantitative validation studies and expert technical review. Full article

This research examines the transformation of film and TV series consumption within the contemporary media landscape, characterized by digital plenitude and Over-the-Top (OTT) dominance. The study investigates how users navigate the transition from linear broadcasting toward on-demand, platform-centric environments. Through an exploratory qualitative approach as the initial phase of a broader study, 18 semi-structured interviews were conducted with a demographically diverse group of Hungarian participants whose primary commonality is active film and TV series consumption. The findings highlight a rejection of traditional linear television, driven by an aversion to intrusive advertising and a demand for temporal autonomy. While mobile devices, particularly smartphones, are central to this shift, consumption remains predominantly stationary; users prioritize the flexibility of cross-device access within the domestic environment over mobile viewing during transit. Furthermore, the study identifies a growing friction caused by content fragmentation between different OTT platforms and rising subscription costs, while digital piracy persists as a marginal alternative. Ultimately, the study concludes that the modern audience acts as a strategic user navigating a complex ecosystem of excess. This underscores a fundamental shift where the cultural value of content is increasingly defined by the tension between individual agency and the systemic constraints of competing services. Full article

The Binglangjiang buffalo, the only indigenous river-type buffalo in China, poses significant challenges for automated keypoint detection due to its uniformly black, low-texture coat, poor foreground–background contrast, and scarcity of annotated training samples. To address these challenges, this study constructs a benchmark dataset of 10,834 lateral-view images covering 424 individuals, annotated with 10 body measurement keypoints following standardized buffalo measurement protocols. A keypoint detection pipeline is developed by adapting DINOv2 with a top-down heatmap regression head under a single-view imaging setup, reducing hardware complexity for practical farm deployment. Benchmarking against YOLOv8 series and a standard ViT baseline shows that DINOv2-Base achieves 96.51% mAP, surpassing YOLOv8m by 5.6 percentage points. Compared to standard ViT, DINOv2 demonstrates more stable localization across keypoints under model scaling. Specifically, on the scapular tip (P8), a particularly low-texture region, DINOv2 exhibits only 0.28% mAP fluctuation versus 0.82% for standard ViT, indicating greater robustness to limited training data and low-contrast imaging. Body measurement validation on 20 individuals yields MAPE values of 1.76–5.69% across five measurements, confirming reliable non-contact measurement performance. The dataset and pipeline provide practical support for precision livestock management of endangered breeds. Full article

Early identification of at-risk students is crucial for timely pedagogical intervention. Determining which assessments instructors should prioritize is complicated by the fact that different eXplainable-AI (XAI) methods can produce conflicting rankings for the same predictive model. We develop a framework combining a sequential GRU model with two complementary XAI techniques, Gradient SHAP (attribution) and DiCE (counterfactuals), and evaluate it in a foundational Data Structures and Algorithms course. The framework produces predictions and explanations for every prefix length throughout the semester and quantifies inter-method agreement and intra-method stability using three disagreement metrics. Intersecting the top-k features identified by both methods isolates a compact subset of assessments whose predictive role is confirmed across two fundamentally different explanation mechanisms. We interpret this cross-method agreement as a heuristic that increases confidence in identified features relative to single-method results, though not as evidence of causal validity. For individual students, the framework uses the intersection of the two types of explanations when it is non-empty; otherwise, the instructor chooses between SHAP’s diagnostic view and DiCE’s prescriptive view, with an optional check against the top-k list. The resulting guidance is less susceptible to method-specific biases than analyses relying on a single method. Full article

Object detection in wide-field-of-view (wide-FoV) unmanned aerial vehicle (UAV) imagery for Internet of Things (IoT) remote sensing applications requires accurate recognition of tiny objects under severe background redundancy and extreme scale variation. As the field of view expands, conventional dense detectors tend to waste substantial computation on non-informative regions, while feature downsampling and static receptive fields often cause the dilution of foreground information and scale confusion. To address these issues, we propose MFRC-Det, a unified framework built upon two complementary principles: mask-guided feature routing and adaptive context modeling. Specifically, a Superpixel-Masking Generator (SP-Masker) is introduced to estimate an image-space soft foreground prior by comparing Simple Linear Iterative Clustering (SLIC) superpixel histograms with a peripheral background reference, propagating the resulting scores on a superpixel adjacency graph, and projecting the refined region-level scores back to a pixel-level routing mask. Guided by these priors, a Greedy-Cutter (G-Cutter) converts dense feature maps into compact, foreground-focused patches without repeated backbone evaluation on cropped image regions, thereby reducing redundant background computation while preserving local structural coherence. On top of the retained regions, an Adaptive Receptive-field Selection Network (ARSNet) aggregates multi-scale contextual responses from several learnable receptive-field candidate branches. ARSNet predicts spatial selection weights conditioned on the input features, allowing each location to emphasize a suitable receptive-field response for object representation. Experimental results on VisDrone-DET and UAVDT demonstrate that MFRC-Det achieves competitive detection accuracy with favorable computational efficiency. Specifically, MFRC-Det obtains 36.1% AP, 60.4% $A{P}_{50}$, and 38.5 FPS on VisDrone-DET and 21.3% AP, 36.8% $A{P}_{50}$, and 37.4 FPS on UAVDT. These results validate the effectiveness of mask-guided feature routing and adaptive context modeling for wide-FoV UAV object detection and suggest their potential value for computation-efficient aerial perception in IoT remote sensing applications. Full article

Fish disease and species identification is critical for intelligent aquaculture, directly influencing productivity, sustainability, and economic viability. However, existing approaches largely treat species identification and pathological classification as independent tasks, limiting their ability to capture interdependent features under complex real-world conditions such as occlusion, low contrast, dynamic backgrounds, and high inter-class similarity. Moreover, challenges including class imbalance, cross-species variability, and fine-grained feature discrimination remain insufficiently addressed. To overcome these limitations, this paper proposes a hybrid ConvNeXt–BiLSTM–multi-head self-attention (MHSA) framework for joint fish species and disease classification, where a ConvNeXt-Small backbone extracts hierarchical spatial features that are transformed into a structured sequence and processed by a bidirectional LSTM to capture contextual dependencies, followed by an MHSA module for adaptive feature refinement. An auxiliary species classification branch is incorporated to provide multi-task regularization without additional inference costs. The training pipeline integrates CLAHE-based image enhancement, square-root inverse-frequency focal loss, targeted minority oversampling, and a two-stage progressive learning strategy with differential-rate cosine annealing, complemented by five-view test-time augmentation. For practical deployment, a YOLOv8s detector is employed for fish localization prior to classification. The experimental results demonstrate that the proposed model achieves superior performance, attaining overall top-1 classification accuracy of 94.33%, precision of 97.1%, recall of 90.9%, 96.1% mAP50, and an F1-score of 0.9264, while achieving a macro AUC of 0.994 and maintaining high computational efficiency (213.3 FPS), demonstrating a robust and efficient solution for real-time fish disease screening. Full article

Detecting hail from remote sensing observations remains challenging because hail develops rapidly and its signatures may appear at different levels within a storm. Ground-based radar and geostationary meteorological satellites are the two primary observing systems for this task, yet their observations are often spatially misaligned. Satellite measurements mainly characterize the thermal structure near the cloud top, whereas radar observations capture the lower-level precipitation core. This mismatch is further exacerbated by satellite parallax, namely the apparent horizontal shift of high cloud tops caused by the oblique viewing geometry of a geostationary satellite, together with the vertical tilt of convective storms. Existing joint methods generally combine satellite cloud-top information with radar precipitation information directly, without explicitly correcting the spatial displacement, which limits detection accuracy. To address this issue, we propose HailDeformer, a deep learning framework that first aligns satellite and radar features through a bidirectional deformable cross-attention module equipped with a position-wise confidence gate and optimized with smoothness, contrastive alignment, and observation-structure consistency losses, and then refines the fused representation using an inter-scale attention module and a wavelet-guided refinement module. Experiments on a four-region dataset from China show that HailDeformer consistently outperforms Direct Fusion, Manual Weighting, Cross-Attention Fusion, and Optical Flow Alignment, achieving a mean Average Precision at IoU 0.5 (mAP@0.5) of 0.916, an F1 score of 0.864, a Critical Success Index (CSI) of 0.760, and the lowest False Alarm Ratio (FAR) of 0.149. Ablation studies further confirm that all proposed modules and associated constraints contribute to the overall performance, with the alignment module providing the largest improvement. Additional evaluations demonstrate that HailDeformer remains effective throughout storm evolution and under challenging observational conditions. Full article

The Laozi Zhigui of Yan Zun occupies an important place in the history of interpretation of the Laozi, and his thoughts had an enormous influence on the xuanxue scholars of the Wei-Jin period afterward. This study examines the ways in which Yan Zun attempted to resolve some contradictions in the theory of the Dao in the Laozi, such as how the Dao relates to the myriad things and to ziran, and the problem of how the Dao can be heard. This study also examines how, in his attempt to resolve these contradictions, Yan Zun could not escape creating a few contradictions of his own, such as the opposition between the theories that the Dao generates things and things self-generate; the opposition between Dao and Virtue following from ziran and the Dao embodying ziran inherently; and how the top, average and lowest kinds of scholars who have heard the Dao, despite all finding joy in their different hearings of the Dao, nevertheless exist in a hierarchical relation to each other. But from the point of view of the Laozi Zhigui as a whole, Yan Zun nevertheless tried to resolve these contradictions. Of course, some of these resolutions succeed, while others do not. This study also examines how, in his effort to resolve these contradictions, Yan Zun often took recourse to Zhuangzi’s conception of the Dao. Full article

This paper presents a dashboard-oriented visual analytics workflow for classification-related exploratory analysis based on Information Gain (IG), histogram segmentation, and complementary localized interpretation through the Precise Piecewise Correlation (PPC) method. The workflow is designed to support the construction of a primary dashboard view by prioritizing attributes with stronger relevance to the decision variable and inspecting their class-related behavior within segmented histogram intervals. Rather than introducing a new standalone feature-selection metric, this study formalizes how established analytical components can be integrated into a coherent dashboard framework for structured visual inspection. The proposed workflow was examined on three datasets from different application domains: the Iris dataset, an educational performance dataset, and an Oil and Gas dataset. Across these cases, IG-based prioritization identified attributes that provided clearer class-related structure in the primary dashboard view, while histogram segmentation supported interval-level interpretation of class concentration and overlap. A compact quantitative evaluation further showed that top-ranked IG subsets retained strong discriminative information under standard classification models, whereas lower-ranked subsets generally performed less favorably. Entropy-based segment analysis additionally indicated lower local class uncertainty for higher-ranked attributes. A small user study provided preliminary user-centered support for the interpretability and practical usefulness of the proposed dashboard structure. The results suggest that the proposed workflow can support dashboard-based inspection of class-related patterns across different contexts. Full article

The acetic acid corrosion resistance of silver electrodes is critical for ensuring photovoltaic (PV) module reliability. Ethylene-vinyl acetate (EVA) is the most widely used encapsulant material in photovoltaic modules. Under exposure to light, heat, and moisture, EVA decomposes to generate acetic acid, which corrodes the silver electrodes, leading to energy conversion efficiency degradation of the module. To address this problem, the lead-free paste was formulated and evaluated in this paper to improve the anti-acetic acid performance. The contact resistivity of the front and the rear side of the solar cells have been measured before and after acetic acid exposure, and greater degradation is shown in the front electrode than in the rear side. Furthermore, the lead-free paste demonstrates lower efficiency degradation compared to the lead-containing paste after acetic acid exposure. In addition, top-view and cross-sectional scanning electron microscopy was performed to analyze the mechanism of the acetic acid corrosion resistance, in which the silver acetate particles were observed. Our experimental results demonstrate that the lead-free paste exhibits superior acetic acid corrosion resistance, which is due to its higher glass acidity and the absence of lead oxide that causes enhanced chemical reactivity with acetic acid. Based on these findings, the acetic acid corrosion model is proposed to attribute the conversion efficiency degradation of reactions between acetic acid and silver, as well as the glass of the silver electrodes. Full article