The automated classification of artistic styles in paintings is a fundamental task in digital art history and computational aesthetics. For years, the dominant paradigm for this image-recognition problem has relied on convolutional neural networks (CNNs), including architectures such as VGG16, VGG19, Inception, EfficientNet, and ResNet, whose inductive biases for local feature extraction have yielded strong performance. However, this reliance on local processing may limit a model's ability to capture global stylistic cues, such as composition, brushwork patterns, and holistic texture, that are often critical for distinguishing artistic movements. To address this limitation, this study proposes AOViT, a Vision Transformer (ViT)-based model tailored to artistic style classification. By exploiting the self-attention mechanism of the ViT architecture, the proposed model captures long-range dependencies and global contextual information. Experimental evaluation on the Persian Miniatures dataset yielded a mean accuracy of 93.85% under five-fold cross-validation. Furthermore, the results indicate that AOViT provides a competitive alternative to CNN-based models for artistic style classification, particularly when combined with appropriate training strategies under limited-data conditions.
IntroductionThe process of measuring the dimensions of a blind hole is difficult because the interior surfaces and base are hard to probe using normal probes and camera equipment. Optical measurement is additionally affected by restricted visibility, lighting differences, reflectance, view angle, and robotic positioning accuracy.MethodsAn adaptive robotic optical metrology system using AI is suggested to achieve non-contact hole inspection in this paper. The system consists of a six DOF robot, an optical measurement head with a camera, adaptive lighting system, AI segmentation, sub-pixel circle and ellipse detection, and robot-camera calibration with error correction using CMM reference data. The system was tested using 30 samples having holes with diameter ranges of 5-15 mm, depth range of 10-45 mm, aspect ratio of 2.0-3.1, three different surface reflectivity conditions, and angle range of 0-20°.ResultsAll trials were successfully carried out for detecting the holes. Average position error in the three-dimensional space was 0.0322 mm; the RMSE of position error was 0.0344 mm; the maximum position error was 0.0645 mm. The average absolute diameter error was 0.0544 mm; RMSE of the absolute diameter error was 0.0647 mm. Average repeatability was 0.0217 mm; average confidence in AI-based segmentation was 0.935. Average position errors varied between 0.026 mm and 0.037 mm depending on reflectivity classes and increased at 15° and higher approach angles.DiscussionThe results offer initial proof-of-concept validation for the use of AI-aided adaptive optical robotic systems to perform automated and contactless inspections of blind holes at least within the current scope of operation. Nevertheless, the small number of samples used, single replication and partially confounding experimental conditions pose a limit on the generalizability of the results. Additional research is necessary using multiple trials, ablation tests of modules, and larger depth-to-diameter ratios are required to establish the system’s robustness and measurement reliability.
Deep learning has become an important computational approach in dermatology because it enables automated segmentation and classification of skin lesions. Manual dermoscopic analysis is time-consuming and depends on specialist expertise, with potential inter-observer variability. There is a need to develop reliable computer-aided diagnostic tools. This structured narrative review covers journal papers published mainly from 2018 to 2026 on skin lesion analysis with focus on image preprocessing, benchmark datasets, conventional machine learning approaches, convolutional neural networks, U-Net architectures, transfer learning approaches, Vision Transformers, attention-based models, ensemble solutions, and hybrid segmentation–classification systems. Building on present reviews covering individual tasks as well as broader segmentation–classification pipelines, the present work provides a critical development-to-deployment analysis connecting dataset characteristics, preprocessing, segmentation, classification, evaluation, explainability, robustness, fairness, computational requirements, and clinical applicability. The paper describes widely used evaluation metrics and explains why results obtained across different datasets, partitions, and experimental setups should not be directly compared. Major open problems include class imbalance, data bias, domain shift, limited representation of diverse skin tones, inadequate external validation, computational requirements, and limited explainability of efficient models. Emerging trends such as self-supervised learning, Transformers, foundation models, explainable AI, federated learning, and multi-modal clinical data are reviewed as well. The review offers practical advice to researchers and clinicians on how to approach the preprocessing, segmentation, classification, and evaluation of skin lesions and highlights priorities in developing clinically reliable skin lesion analysis systems.
The classification of research papers from different academic data is still challenging due to the presence of semantic ambiguity, lack of information, fluctuating credibility, and complex linkages across research articles. Existing fuzzy graph neural networks depend too much on membership information and cannot truly consider the intuitionistic uncertainty, trust dependability, and structural importance of nodes at the graph propagation level. To overcome these constraints, this paper presents an intuitionistic fuzzy graph neural network architecture that combines trust-aware intuitionistic representations and a unique node Laplacian energy mechanism. The suggested Node Laplacian Energy is derived from the graph Laplacian's row-wise norm and is used as an adaptive node-weighting metric to assess each research paper's local structural contribution. Furthermore, a diffusion-based propagation technique is used to enhance information transmission across semantically linked publications and capture larger graph relationships. The suggested system is tested for research-paper categorization with semantic similarity features and intuitionistic fuzzy credibility information. Its performance is compared to known machine learning and graph neural network baselines utilizing accuracy, precision, recall, and F1-score. The findings show that including uncertainty-aware representations, structural node weighting, and diffusion-based graph propagation can improve discriminative academic content categorization. The suggested paradigm offers a unified way to model semantic linkages, uncertainty, believability, and graph structural relevance in research-paper categorization.
