Abstract
High-resolution UAV imagery offers unprecedented opportunities for vineyard monitoring, yet its practical use in semantic segmentation is hindered by the high cost of pixel-level annotation. Weakly supervised learning (WSL) emerges as a promising alternative, capable of reducing annotation effort while preserving competitive performance. In this study, we conduct a direct comparative evaluation of two pseudo-labelling strategies for vine row segmentation, a task still underexplored in perennial crops. The first strategy combines a spectral heuristic with Conditional Random Fields (CRF) to enforce spatial consistency, while the second employs token clustering of DINO-ViT embeddings. To ensure fairness, both pseudo-label sets were used to train an identical segmentation architecture (U-Net with ResNet50), thereby isolating the impact of pseudo-label quality. Results, measured by precision, recall, F1-score, and Intersection over Union (IoU), reveal that the CRF-refined heuristic (F1 = 0.77, IoU = 0.62) consistently outperforms the transformer-based clustering approach (F1 = 0.52, IoU = 0.50). These findings highlight the decisive role of spatial regularisation in weak supervision and provide a reproducible pipeline that balances accuracy, methodological simplicity, and annotation cost. The contribution of this work lies in demonstrating a practical and extensible framework for UAV-based vineyard monitoring, while opening pathways for hybrid approaches that integrate semantic depth with spatial coherence in future research.

Abstract
This article presents a Ph.D. research proposal for the automation of Digital Twin construction in industrial contexts through the semantic integration of heterogeneous data. The approach combines Large Language Model with the Asset Administration Shell framework to extract and map technical information from structured and unstructured sources (such as sensors, manuals and ERP/MES systems) into standardized submodels. The methodology includes four stages: data collection, semantic mapping using, organization into submodels and integration into Digital Twins. Initial tests with simulated data show the ability of LLMs to identify equivalent technical terms and generate structured data compatible with Asset Administration Shell. Ongoing work includes future activities with data from industrial partners, development of evaluation metrics and analysis with domain experts. The aim is to reduce manual modeling work, support interoperability and enable the construction of scalable Digital Twin in line with Industry 4.0 frameworks.

Abstract
Traffic safety is a significant issue, particularly for motorcyclists, who are significantly more vulnerable to fatal accidents compared to drivers of enclosed vehicles. Advanced Rider Assistance Systems have the potential to improve motorcyclist safety by mitigating critical risks like rear-end collisions. This paper proposes an integrated Advanced Rider Assistance Systems architecture using the RTDETR object detection model to predict rear-end collisions and provide real-time alerts to the riders. Specifically, it aims to mitigate critical risks like rear-end collisions, offering a comprehensive safety solution. The experimental data indicate that the proposed system is suitable for diverse operational environments, supporting the development of advanced sensing and alert systems to improve motorcyclist safety. Our system achieves an average precision (AP) of 0.688 at IoU=0.50 and reduces collision risk by issuing timely warnings. Contributions include the integration of RTDETR for improved detection accuracy, a multi-threshold warning mechanism, and a detailed analysis of system performance under varied conditions.

Abstract
Video capsule endoscopy (VCE) enables high-resolution visualisation of the small bowel but remains constrained by manual review of thousands of frames, which is time-consuming and error-prone under class imbalance. This study investigates deep learning for automatic multiclass lesion classification in VCE, comparing two convolutional networks (ResNet-50, EfficientNet-B3) with two Vision Transformers (Swin, DeiT) on the public Kvasir-Capsule dataset (47,161 images; 11 classes). The pipeline comprises standard preprocessing, class-aware augmentation and adaptive data augmentation, stratified data partitioning, hyperparameter optimisation with Optuna, and evaluation using accuracy, precision, recall, and F1-score. DeiT achieved the best overall performance (accuracy = 0.98; F1 = 0.96), with strong class-wise results in clinically salient categories (e.g., ulcer, fresh blood, angiectasia), indicating effective modelling of long-range dependencies and subtle patterns. We further assess computational feasibility by reporting training configuration and indicative inference time per image, supporting potential integration into assisted reading workflows. Limitations include reliance on a single public dataset, pronounced class imbalance, and the absence of prospective clinical validation, which may affect generalisability. These findings position Transformer-based models as promising candidates for VCE decision support, while underscoring the need for future work on (i) multicentric datasets and external validation, (ii) comprehensive statistical analysis with confidence intervals and robust baselines under imbalance, and (iii) prospective studies quantifying end-to-end impact on reading time and diagnostic safety. © 2025 The Authors. Published by Elsevier B.V.

Abstract
Autonomous docking remains a critical maneuver for the continuous operation of Unmanned Surface Vehicles (USVs) in challenging marine environments. This paper proposes a docking strategy for a non-holonomic 2-DoF vessel. Because the vehicle is strictly limited to surge and yaw actuation-lacking lateral sway capabilities-the terminal docking phase becomes a highly constrained maneuver. To overcome this, our approach relies on Linear Parameter-Varying Model Predictive Control (LPV-MPC) coupled with Dubins Path trajectory generation. While MPC offers predictive capabilities and constraint handling, standard point-regulation approaches under such strict underactuation often result in oscillatory behavior and inefficient actuation during the final approach. To address this, a guidance layer that generates smooth, curvature-constrained reference trajectories was proposed to ensure kinematic feasibility before control optimization. Furthermore, a comparative analysis of the results from both approaches was conducted under strict actuator constraints. © 2026 IEEE.

Abstract
The increasing connectivity of Internet of Medical Things (IoMT) devices has accentuated their susceptibility to cyberattacks. The sensitive data they handle makes them prime targets for information theft and extortion, while outdated and insecure communication protocols further elevate security risks. This paper presents a lightweight and innovative approach that combines Benford's law with statistical distance functions to detect attacks in IoMT devices. The methodology uses Benford's law to analyze digit frequency and classify IoMT devices traffic as benign or malicious, regardless of attack type. It employs distance-based statistical functions like Jensen-Shannon divergence, KullbackLeibler divergence, Pearson correlation, and the Kolmogorov test to detect anomalies. Experimental validation was conducted on the CIC-IoMT-2024 benchmark dataset, comprising 45 features and multiple attack types. The best performance was achieved with the Kolmogorov test (alpha = 0.01), particularly in DoS ICMP attacks, yielding a precision of.99.24%, a recall of.98.73%, an F1 score of.98.97%, and an accuracy of.97.81%. Jensen-Shannon divergence also performed robustly in detecting SYN-based attacks, demonstrating strong detection with minimal computational cost. These findings confirm that Benford's law, when combined with well-chosen statistical distances, offers a viable and efficient alternative to machine learning models for anomaly detection in constrained environments like IoMT.