Fatema A. Albalooshi

Dr. Fatema A. Albalooshi

Computer Engineering, University of Bahrain, Madīnat Ḩamad, Ash Shamālīyah, Bahrain
orcid

Research Focus & Metrics

Academic Career Length

15Years (2012-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
29 July 2018 in Microprocessors and Microsystems
Verified
Journal Article
Closed access
MSMax Smith-Creasey
Max Smith-Creasey
Fatema A. AlbalooshiFatema A. Albalooshi
Fatema A. Albalooshi
MRMuttukrishnan Rajarajan
Muttukrishnan Rajarajan

30 January 2025 in Applied Sciences
Verified
Journal Article
Open access
PDF file
Fatema A. AlbalooshiFatema A. Albalooshi
Fatema A. Albalooshi
MQM. R. Qader
M. R. Qader
Power quality disturbances (PQDs) are major obstacles to maintaining the reliability and stability of electrical systems. This study introduces a new multi-scale deep learning method to classify PQDs, aiming to enhance the accuracy and efficiency of power quality (PQ) analysis and monitoring systems. By combining 1-D convolutional neural networks (CNNs) with an attention mechanism, this approach overcomes the limitations of traditional techniques. Moreover, varying-size convolutional layers allow for the direct learning of complex patterns and features from PQ signals. To address the challenge of limited labeled PQ datasets, this research utilizes an open-source dataset generator to create large-scale datasets with annotated PQDs. Through a comparison with existing models in the field, the superiority of the proposed CNN-based approach is evident, achieving an accuracy level of up to 99.49%. The results demonstrate promising classification performance in terms of simplicity and accuracy, highlighting the potential of this approach to improve PQ analysis and disturbance identification.

8 November 2024 in Applied Sciences
Verified
Journal Article
Open access
PDF file
Fatema A. AlbalooshiFatema A. Albalooshi
Fatema A. Albalooshi
Vegetation segmentation plays a crucial role in accurately monitoring and analyzing vegetation cover, growth patterns, and changes over time, which in turn contributes to environmental studies, land management, and assessing the impact of climate change. This study explores the potential of a multi-scale convolutional neural network (MSCNN) design for object classification, specifically focusing on vegetation detection. The MSCNN is designed to integrate multi-scale feature extraction and attention mechanisms, enabling the model to capture both fine and coarse vegetation patterns effectively. Moreover, the MSCNN architecture integrates multiple convolutional layers with varying kernel sizes (3 × 3, 5 × 5, and 7 × 7), enabling the model to extract features at different scales, which is vital for identifying diverse vegetation patterns across various landscapes. Vegetation detection is demonstrated using three diverse datasets: the CamVid dataset, the FloodNet dataset, and the multispectral RIT-18 dataset. These datasets present a range of challenges, including variations in illumination, the presence of shadows, occlusion, scale differences, and cluttered backgrounds, which are common in real-world scenarios. The MSCNN architecture allows for the integration of information from multiple scales, facilitating the detection of diverse vegetation types under varying conditions. The performance of the proposed MSCNN method is rigorously evaluated and compared against state-of-the-art techniques in the field. Comprehensive experiments showcase the effectiveness of the approach, highlighting its robustness in accurately segmenting and classifying vegetation even in complex environments. The results indicate that the MSCNN design significantly outperforms traditional methods, achieving a remarkable global accuracy and boundary F1 score (BF score) of up to 98%. This superior performance underscores the MSCNN’s capability to enhance vegetation detection in imagery, making it a promising tool for applications in environmental monitoring and land use management.

24 December 2024 in Applied Sciences
Verified
Journal Article
Open access
PDF file
EEEbenezer Kojo Essel
Ebenezer Kojo Essel
FLFred Lacy
Fred Lacy
Fatema A. AlbalooshiFatema A. Albalooshi
Fatema A. Albalooshi
WEWael Elmedany
Wael Elmedany
YIYasser Ismail
Yasser Ismail
Drowsiness has been recognized as a leading factor in road accidents worldwide. Despite considerable research in this area, this paper aims to improve the precision of drowsiness detection specifically for long-haul travel by employing the Dlib-based facial feature detection algorithm. This study proposes two algorithms: a static and adaptive frame threshold. Both approaches utilize eye closure ratio (ECR) and mouth aperture ratio (MAR) parameters to determine the driver’s level of drowsiness. The static threshold method issues a warning when the ECR and/or MAR values reach specific thresholds. In this method, the ECR threshold is established at 0.15 and the MAR threshold at 0.4. The static threshold method demonstrated an accuracy of 89.4% and a sensitivity of 96.5% using 1000 images. The adaptive frame threshold algorithm uses a counter to monitor the number of consecutive frames that meet the drowsiness criteria before triggering a warning. Additionally, the number of consecutive frames required is adjusted dynamically over time to enhance detection accuracy and more accurately indicate a state of drowsiness. The adaptive frame threshold algorithm was tested using four 30 min videos, from a publicly available dataset achieving a maximum accuracy of 98.2% and a sensitivity of 64.3% with 500 images.

2 October 2025 in Future Transportation
Verified
Journal Article
Open access
PDF file
Fatema A. AlbalooshiFatema A. Albalooshi
Fatema A. Albalooshi
The accelerating pace of urbanization has significantly complicated traffic management systems, leading to mounting challenges, such as persistent congestion, increased travel delays, and heightened environmental impacts. In response to these challenges, this study presents a novel deep learning framework designed to enhance short-term traffic flow prediction and support intelligent transportation systems within the context of smart cities. The proposed model integrates Gated Recurrent Units (GRUs) and Long Short-Term Memory (LSTM) networks, augmented by an attention mechanism that dynamically emphasizes relevant temporal patterns. The model was rigorously evaluated using the publicly available datasets and demonstrated substantial improvements over current state-of-the-art methods. Specifically, the proposed framework achieves a 3.75% reduction in the Mean Absolute Error (MAE), a 2.00% reduction in the Root Mean Squared Error (RMSE), and a 4.17% reduction in the Mean Absolute Percentage Error (MAPE) compared to the baseline models. The enhanced predictive accuracy and computational efficiency offer significant benefits for intelligent traffic control, dynamic route planning, and proactive congestion management, thereby contributing to the development of more sustainable and efficient urban mobility systems.

1 August 2018
Verified
Conference Paper
Closed access
MSMax Smith-Creasey
Max Smith-Creasey
Fatema A. AlbalooshiFatema A. Albalooshi
Fatema A. Albalooshi
MRMuttukrishnan Rajarajan
Muttukrishnan Rajarajan
In this paper we produce a transparent continuous facial authentication scheme that is aware of the current activity and illumination context. It has recently been shown that contextual differences in environment and user activity have significant effects on the accuracy of the scheme. Furthermore, cross-comparisons or permutations of facial data captured in different contexts can see detrimental authentication results. Our scheme models both ambient light and activity accelerometer data for contextual awareness. This data is used to train separate facial classifiers for different activity and illumination contexts (e.g.: when a user is walking). When a face is captured for classification, a window of accelerometer and ambient light data is firstly classified to select which classifier has been trained on facial data obtained from the current context. In our experiments we use two state-of-the-art facial datasets. We describe the architecture and performance of our context recognition components. We show that activity and illumination awareness decreases the equal error rate by 4.05% and 4.39%, respectively. We also perform experiments to show the required accuracy needed from contextual components to yield facial recognition improvements.