Abstract
Motorcycle safety is significantly compromised by blind spot collisions, necessitating Advanced Rider Assistance Systems. This paper proposes a blind spot warning system designed for edge devices, leveraging computer vision techniques for real-time object detection and tracking. The system aims to enhance rider awareness by detecting vehicles in blind spots and providing timely warnings through visual LED alerts. Preliminary results from training the RF-DETRBase model, the lightest version of the RF-DETR architecture, on the challenging BDD100K dataset, which features diverse driving scenarios and numerous small objects, demonstrate the system's potential. The dataset's inherent complexities were highlighted by poor small object detection (mAP 0.128) and declining performance at higher IoU thresholds. Despite these challenges, the model achieved a promising Average Precision of 0.700 at an Intersection over Union of 0.50, indicating effective vehicle detection.

Abstract
One of the key aspects of Industry 4.0 is using intelligent systems to optimize manufacturing processes by improving productivity and reducing costs. These systems have greatly impacted in different areas, such as demand prediction and quality assessment. However, the prognostics and health management of industrial equipment is one of the areas with greater potential. This paper presents a comparative analysis of deep learning architectures applied to the prediction of the remaining useful life (RUL) on public real industrial datasets. The analysis includes some of the most commonly employed recurrent neural network variations and a novel approach based on a hybrid architecture using transformers. Moreover, we apply explainability techniques to provide comprehensive insights into the model's decision-making process. The contributions of the work are: (1) a novel transformer-based architecture for RUL prediction that outperforms traditional recurrent neural networks; (2) a detailed description of the design strategies used to construct the models on two under-explored datasets; (3) the use of explainability techniques to understand the feature importance and to explain the model's prediction and (4) making models built for reproducibility available to other researchers.

Abstract
Retinal diseases such as glaucoma, diabetic retinopathy, and age-related macular degeneration affect hundreds of millions of people worldwide and are among the leading causes of vision loss. Optical Coherence Tomography (OCT) is a non-invasive imaging technique widely used to support the diagnosis of these conditions. However, manual analysis of OCT images is time-consuming, prone to inter-observer variability, and requires extensive clinical expertise. In recent years, deep learning methods have shown outstanding performance in medical image segmentation tasks. This work proposes an automatic approach for the segmentation of retinal layers in OCT images using the GOALS 2022 dataset. Four segmentation architectures were evaluated - U-Net, DeepLabV3+, FPN (U-Net++), and Attention U-Net - all combined with the ResNet50 encoder. Additionally, the influence of encoder selection in the U-Net architecture was investigated, testing ResNet34, EfficientNetB0, MobileNetV2, VGG16, and InceptionV3. The results show that the DeepLabV3+ model achieved the best overall performance, with an F1-Score of 0.9669 and an IoU of 0.9370. These findings demonstrate that lightweight, accessible models can achieve results comparable to state-of-the-art methods, offering a promising solution for clinical applications in retinal image segmentation. © 2025 The Authors. Published by Elsevier B.V.

Abstract
Motorcycle accidents contribute significantly to road traffic fatalities worldwide. Advanced Rider Assistance Systems are systems designed to improve safety by providing real-time monitoring, collision detection, and adaptive control. However, implementing these systems on low-power hardware poses some challenges. This work examines the usage of object detection models on edge computing devices, with a focus on the Raspberry Pi 4. In this work are evaluated some models like SSD, and YOLO models using a custom dataset based on BDD100K dataset. All the models used in the project were converted to Open Neural Network Exchange format and tested at resolutions of 320 x 320 and 640 x 640 to assess their efficiency and real-time applicability. The results indicate that YOLO-based models, particularly YOLOv11, give the best balance between accuracy and inference speed, making them a candidate for Advanced Rider Assistance Systems applications. Despite these advancements, challenges such as real-time constraints and hardware limitations remain. Future research should focus on model quantization and hardware acceleration to improve deployment feasibility.

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. © 2025 The Authors. Published by Elsevier B.V.

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.