arXiv subjects

Omar Awadallah

Publications and source records attributed to Omar Awadallah.

2 recordsLinked to original sources

arXiv2025

Augmented Reality-Based Smart Structural Health Monitoring System With Accurate 3D Model Alignment

Structural Health Monitoring (SHM) has become increasingly critical due to the rapid deterioration of civil infrastructure. Traditional methods involving heavy equipment are costly and time-consuming. Recent SHM approaches use advanced non-contact sensors, IoT, and Augmented Reality (AR) glasses for faster inspections and immersive experiences during inspections. However, current methods lack quantitative damage data, remote collaboration support, and accurate 3D model alignment with the real structure. Recognizing these current challenges, this paper proposes an AR-based system that integrates Building Information Modelling (BIM) visualization and follows a flexible manipulation approach of 3D holograms to improve structural condition assessments. The proposed framework utilizes the Vuforia software development toolkit to enable the automatic alignment of 3D models to the real structure, ensuring successful model alignment to assist users in accurately visualizing damage locations. The framework also enables flexible manipulation of damage locations, making it easier for users to identify multiple damage points in the 3D models. The system is validated through lab-scale and full-scale bridge use cases, with data transfer performance analyzed under 4G and 5G conditions for remote collaboration. This study demonstrates that the proposed AR-based SHM framework successfully aligns 3D models with real structures, allowing users to manually adjust models and damage locations. The experimental results confirm its feasibility for remote collaborative inspections, highlighting significant improvements with 5G networks. Nevertheless, performance under 4G remains acceptable, ensuring reliability even without 5G coverage.

cs.HC↗

arXiv2025

Any-Class Presence Likelihood for Robust Multi-Label Classification with Abundant Negative Data

Multi-label Classification (MLC) assigns an instance to one or more non-exclusive classes. A challenge arises when the dataset contains a large proportion of instances with no assigned class, referred to as negative data, which can overwhelm the learning process and hinder the accurate identification and classification of positive instances. Nevertheless, it is common in MLC applications such as industrial defect detection, agricultural disease identification, and healthcare diagnosis to encounter large amounts of negative data. Assigning a separate negative class to these instances further complicates the learning objective and introduces unnecessary redundancies. To address this challenge, we redesign standard MLC loss functions by deriving a likelihood of any class being present, formulated by a normalized weighted geometric mean of the predicted class probabilities. We introduce a regularization parameter that controls the relative contribution of the absent class probabilities to the any-class presence likelihood in positive instances. The any-class presence likelihood complements the multi-label learning by encouraging the network to become more aware of implicit positive instances and improve the label classification within those positive instances. Experiments on large-scale datasets with negative data: SewerML, modified COCO, and ChestX-ray14, across various networks and base loss functions show that our loss functions consistently improve MLC performance of their standard loss counterparts, achieving gains of up to 6.01 percentage points in F1, 8.06 in F2, and 3.11 in mean average precision, all without additional parameters or computational complexity. Code available at: https://github.com/ML-for-Sensor-Data-Western/gmean-mlc

cs.LG↗