Yazar "Mowla, Md. Najmul" seçeneğine göre listele

Listeleniyor 1 - 9 / 9
  • [ X ]
    Öğe
    A Lightweight Deep Learning Model for Retinopathy of Prematurity Classification in eHealth Applications
    (IEEE, 2025) Mowla, Neazmul; Mowla, Md. Najmul; Rabie, Khaled; Alsinglawi, Belal
    Retinopathy of Prematurity (ROP) is a vision-threatening condition in premature infants requiring timely and accurate diagnosis to prevent blindness. While electronic health (eHealth) technologies promise to improve neonatal care, automating ROP diagnosis faces challenges such as limited labeled datasets, architectural complexity, and high computational demands. This study introduces LightEyeNet, a lightweight deep-learning architecture optimized for eHealth applications in ROP severity classification. By integrating DenseNet121 and a channel-wise residual attention network block, LightEyeNet enhances diagnostic accuracy and efficiency. Explainable AI techniques, including Grad-CAM and LIME, further improve transparency and clinical interpretability. LightEyeNet achieves 96.28% testing accuracy, outperforming state-of-the-art pre-trained networks, including DenseNet201 (95.78%, + 0.5%), Inception-V3 (93.80%, + 2.48%), Xception (94.54%, + 1.74%), and EfficientDense (87.10%, + 9.18%). Furthermore, LightEyeNet is the most compact architecture among these, with a size of 2.45 MB, compared to Efficient-Dense (6.88 MB), Inception-V3 (3.56 MB), Xception (21.48 MB), and DenseNet201 (5.05 MB). With a specificity of 0.99, a sensitivity of 0.95, and an AUC score of 0.99 across five ROP severity classes, LightEyeNet demonstrates a balance of superior performance and efficiency.
  • [ X ]
    Öğe
    Adaptive Hierarchical Multi-Headed Convolutional Neural Network With Modified Convolutional Block Attention for Aerial Forest Fire Detection
    (IEEE-Inst Electrical Electronics Engineers Inc, 2025) Mowla, Md. Najmul; Asadi, Davood; Masum, Shamsul; Rabie, Khaled
    Effective detection and classification of forest fire imagery are critical for timely and efficient wildfire management. Convolutional Neural Networks (CNNs) have demonstrated potential in this domain but encounter limitations when addressing varying scales, resolutions, and complex spatial dependencies inherent in wildfire datasets. Building upon our prior work on the Unmanned Aerial Vehicle-based Forest Fire Database (UAVs-FFDB) and the multi-headed CNN (MHCNN), this study introduces a novel architecture, namely, the Adaptive Hierarchical Multi-Headed Convolutional Neural Network with Modified Convolutional Block Attention Module (AHMHCNN-mCBAM). This enhanced framework addresses prior challenges by integrating adaptive pooling, concatenated convolutions for multi-scale feature extraction, and an improved attention mechanism incorporating shared fully connected layers, Glorot initialization, rectified linear units (ReLU), layer normalization, and attention-gating. AHMHCNN-mCBAM incorporates Gated Recurrent Unit (GRU) and Bidirectional Long Short-Term Memory (BiLSTM) networks for temporal context modeling to further refine classification accuracy. Experiments conducted on the UAVs-FFDB dataset achieved outstanding results, including 100% accuracy, a 100% Cohen's kappa coefficient (cKappa), and compact model parameter sizes of 1.49 million (M), 0.25 M, and 0.12 M. On the Fire Luminosity Airborne-based Machine Learning Evaluation (FLAME) dataset, the model attained accuracy rates of 99.83%, 99.10%, and 99.32%, with corresponding cKappa values of 99.66%, 98.20%, and 98.65%. Compared to the baseline hierarchical MHCNN with CBAM (HMHCNN-CBAM), AHMHCNN-mCBAM demonstrated significant performance gains, including a 6.80% and 6.59% increase in accuracy, a 9.26% and 14.11% improvement in cKappa, and a 13.87% and 13.76% reduction in parameter size on the UAVs-FFDB and FLAME datasets, respectively. Additionally, AHMHCNN-mCBAM outperformed HMHCNN-CBAM in recall (25% improvement), precision (21.95%), F1-score (14.94%), and fire detection rate (FDR) reduction (25.01%), while achieving a 100% reduction in error warning rate (EWR). Leveraging Explainable Artificial Intelligence (XAI) techniques, the model provides interpretable insights into decision-making processes.
  • [ X ]
    Öğe
    Experimental motor fault detection and identification of a quadrotor UAV using a hybrid deep learning approach
    (Springer Nature, 2025) Khaneghaei, Mohammad; Asadi, Davood; Mowla, Md. Najmul; Disken, Gokay
    This study presents a novel experimental hybrid sequential deep learning (DL) approach for real-time motor fault detection and magnitude estimation in quadrotor UAVs, addressing critical gaps in current fault-tolerant control systems. The proposed framework integrates long short-term memory (LSTM) networks with 1D convolutional neural networks (1D-CNN) to enhance fault classification and estimation accuracy. The dual capability distinguishes the proposed model from existing methods, which often focus solely on fault detection without addressing magnitude estimation. A novel dataset, generated through Hardware-in-the-Loop (HIL) experiments, incorporates 25,000 unique fault scenarios under diverse configurations and flight conditions. This dataset strengthens the model's robustness and applicability to real-world scenarios. The developed architecture demonstrated superior performance, achieving 99.2% fault detection accuracy, surpassing existing methods in robustness and efficiency. By embedding the model into a dynamic HIL testbed, the study validates the framework's capability to detect faults, estimate magnitudes, and restore stability in quadrotors under challenging conditions. Experimental results highlight the system's effectiveness in reducing motor anomalies, ensuring improved operational safety and reliability. The approach is adaptable to broader UAV systems, offering significant advancements in autonomous fault-tolerant control.
  • [ X ]
    Öğe
    Internet of Things and Wireless Sensor Networks for Smart Agriculture Applications: A Survey
    (IEEE-Inst Electrical Electronics Engineers Inc, 2023) Mowla, Md. Najmul; Mowla, Neazmul; Shah, A. F. M. Shahen; Rabie, Khaled M.; Shongwe, Thokozani
    The increasing food scarcity necessitates sustainable agriculture achieved through automation to meet the growing demand. Integrating the Internet of Things (IoT) and Wireless Sensor Networks (WSNs) is crucial in enhancing food production across various agricultural domains, encompassing irrigation, soil moisture monitoring, fertilizer optimization and control, early-stage pest and crop disease management, and energy conservation. Wireless application protocols such as ZigBee, WiFi, SigFox, and LoRaWAN are commonly employed to collect real-time data for monitoring purposes. Embracing advanced technology is imperative to ensure efficient annual production. Therefore, this study emphasizes a comprehensive, future-oriented approach, delving into IoT-WSNs, wireless network protocols, and their applications in agriculture since 2019. It thoroughly discusses the overview of IoT and WSNs, encompassing their architectures and summarization of network protocols. Furthermore, the study addresses recent issues and challenges related to IoT-WSNs and proposes mitigation strategies. It provides clear recommendations for the future, emphasizing the integration of advanced technology aiming to contribute to the future development of smart agriculture systems.
  • [ X ]
    Öğe
    Long-term projections of global, northern hemisphere, and arctic sea ice concentration using statistical and deep learning approaches
    (Pergamon-Elsevier Science Ltd, 2025) Bilgili, Mehmet; Pinar, Engin; Mowla, Md. Najmul; Durhasan, Tahir; Aksoy, Muhammed M.
    The accelerating decline in sea ice concentration (SIC) poses significant challenges for global climate regulation, maritime navigation, and arctic ecosystem stability. This study develops and evaluates two advanced time-series forecasting models, seasonal autoregressive integrated moving average (SARIMA) and long short-term memory (LSTM) networks, to project SIC trends through 2050 across three spatial domains: the globe, the northern hemisphere, and the arctic. Utilizing the ERA5 reanalysis dataset (1970-2024) from the European center for medium-range weather forecasts (ECMWF), the models capture seasonal cycles and complex temporal dependencies to enable robust long-term projections. Comparative analysis demonstrates that SARIMA effectively models periodic fluctuations, while LSTM excels at learning nonlinear dependencies inherent in SIC dynamics. Performance metrics, including mean absolute percentage error (MAPE), root mean square error (RMSE), and correlation coefficient (R), confirm the high accuracy of both models, with SARIMA showing superior capability in representing structured seasonal patterns. Projections indicate a persistent decline in SIC, with arctic concentrations decreasing from 55.60% in 2023 to approximately 46.84% by 2050, underscoring the pronounced effects of arctic amplification. These results provide valuable insights for climate modeling, arctic policy formulation, and the development of adaptive navigation strategies in a rapidly changing polar environment.
  • [ X ]
    Öğe
    Multimodal deep learning for estimating mean precipitable water vapor
    (Wiley, 2026) Mowla, Md. Najmul; Aksoy, Muhammed M.; Pinar, Engin; Bilgili, Mehmet; Durhasan, Tahir
    Precipitable water vapor (PWV) is a crucial atmospheric variable that influences weather systems, climate variability, and hydrological processes. Accurate PWV estimation is essential for improving numerical weather prediction, climate modeling, and remote-sensing applications. However, existing methods often rely on extensive meteorological inputs or computationally intensive architectures, limiting their applicability in data-sparse regions. This study introduces a novel hybrid framework, EMMA-NN-BiGRU-XGBoost, designed to forecast monthly mean PWV across Turkey using only four physically meaningful inputs: latitude, longitude, altitude, and seasonal indicators. The framework integrates an enhanced multimodal attention (EMMA) mechanism that disentangles spatial, altitudinal, and seasonal influences, improving interpretability and physical consistency. Bidirectional gated recurrent units (BiGRU) capture temporal dependencies, and XGBoost models nonlinear feature interactions within a weighted stacking ensemble. Hyperparameters are optimized via particle swarm optimization and Bayesian optimization, with particle swarm optimization demonstrating superior tuning efficiency. Extensive benchmarking against traditional machine-learning models, using grid search and random search with fivefold cross-validation, as well as deep-learning baselines, demonstrates significant improvements in predictive accuracy, achieving an root-mean-square error of and an of 0.92, representing a 15%-20% reduction in error compared with state-of-the-art methods. The model also exhibits robustness across diverse climatic zones in Turkey. Shapley additive explanations further elucidate feature importance, aligning model outputs with climatological principles. Beyond methodological advances, this work provides a scalable, interpretable, and data-efficient baseline for PWV forecasting, thereby facilitating enhanced climate diagnostics, hydrological risk assessments, and early warning systems, particularly in regions with limited meteorological observations.
  • [ X ]
    Öğe
    Real-time fault detection in multirotor UAVs using lightweight deep learning and high-fidelity simulation data with single and double fault magnitudes
    (Springer Heidelberg, 2025) Mowla, Md. Najmul; Asadi, Davood; Sohel, Ferdous
    Robust fault detection and diagnosis (FDD) in multirotor unmanned aerial vehicles (UAVs) remains challenging due to limited actuator redundancy, nonlinear dynamics, and environmental disturbances. This work introduces two lightweight deep learning architectures: the Convolutional-LSTM Fault Detection Network (CLFDNet), which combines multi-scale one-dimensional convolutional neural networks (1D-CNN), long short-term memory (LSTM) units, and an adaptive attention mechanism for spatio-temporal fault feature extraction; and the Autoencoder LSTM Multi-loss Fusion Network (AELMFNet), a soft attention-enhanced LSTM autoencoder optimized via multi-loss fusion for fine-grained fault severity estimation. Both models are trained and evaluated on UAV-Fault Magnitude V1, a high-fidelity simulation dataset containing 114,230 labeled samples with motor degradation levels ranging from 5% to 40% in the take-off, hover, navigation, and descent phases, representing the most probable and recoverable fault scenarios in quadrotor UAVs. Including coupled faults enables models to learn correlated degradation patterns and actuator interactions while maintaining controllability under standard flight laws. CLFDNet achieves 96.81% precision in fault severity classification and 100% accuracy in motor fault localization with only 19.6K parameters, demonstrating suitability for real-time onboard applications. AELMFNet achieves the lowest reconstruction loss of 0.001 with Huber loss and an inference latency of 6 ms/step, underscoring its efficiency for embedded deployment. Comparative experiments against 15 baselines, including five classical machine learning models, five state-of-the-art fault detection methods, and five attention-based deep learning variants, validate the effectiveness of the proposed architectures. These findings confirm that lightweight deep models enable accurate and efficient diagnosis of UAV faults with minimal sensing.
  • [ X ]
    Öğe
    Recent advancements in morphing applications: Architecture, artificial intelligence integration, challenges, and future trends-a comprehensive survey
    (Elsevier France-Editions Scientifiques Medicales Elsevier, 2025) Mowla, Md. Najmul; Asadi, Davood; Durhasan, Tahir; Jafari, Javad Rashid; Amoozgar, Mohammadreza
    This study provides a comprehensive review of recent advancements in aerospace morphing technologies, focusing on integrating artificial intelligence (AI) into morphing architectures. It emphasizes AI's pivotal role in optimizing these systems, particularly through machine learning (ML), deep learning (DL), and reinforcement learning (RL), to enhance real-time adaptability, performance, and efficiency. The review categorizes developments in smart materials, compliant mechanisms, and adaptive structures, offering a detailed analysis of their architectural foundations. It further examines AI-driven aerodynamic optimization and control systems, highlighting recent solutions to structural integrity, energy efficiency, and scalability challenges. Key contributions since 2020 are synthesized through a year-by-year analysis, offering a clear overview of the research landscape. The paper also addresses emerging challenges in aerospace morphing and proposes strategies to alleviate them. Recommendations for future advancements emphasize the integration of state-of-the-art technologies. By critically evaluating current capabilities and limitations, this review provides valuable insights for researchers and practitioners, identifying AI's transformative potential in morphing systems and outlining the technical challenges that must be addressed for future morphing aerospace applications.
  • [ X ]
    Öğe
    UAVs-FFDB: A high-resolution dataset for advancing forest fire detection and monitoring using unmanned aerial vehicles (UAVs)
    (Elsevier Inc., 2024) Mowla, Md. Najmul; Asadi, Davood; Tekeoglu, Kadriye Nur; Masum, Shamsul; Rabie, Khaled
    Forest ecosystems face increasing wildfire threats, demanding prompt and precise detection methods to ensure efficient fire control. However, real-time forest fire data accessibility and timeliness require improvement. Our study addresses the challenge through the introduction of the Unmanned Aerial Vehicles (UAVs) based forest fire database (UAVs-FFDB), characterized by a dual composition. Firstly, it encompasses a collection of 1653 high-resolution RGB raw images meticulously captured utilizing a standard S500 quadcopter frame in conjunction with a RaspiCamV2 camera. Secondly, the database incorporates augmented data, culminating in a total of 15560 images, thereby enhancing the diversity and comprehensiveness of the dataset. These images were captured within a forested area adjacent to Adana Alparslan Türkeş Science and Technology University in Adana, Turkey. Each raw image in the dataset spans dimensions from 353 × 314 to 640 × 480, while augmented data ranges from 398 × 358 to 640 × 480, resulting in a total dataset size of 692 MB for the raw data subset. In contrast, the augmented data subset accounts for a considerably larger size, totaling 6.76 GB. The raw images are obtained during a UAV surveillance mission, with the camera precisely angled a -180-degree to be horizontal to the ground. The images are taken from altitudes alternating between 5 - 15 meters to diversify the field of vision and to build a more inclusive database. During the surveillance operation, the UAV speed is 2 m/s on average. Following this, the dataset underwent meticulous annotation using the advanced annotation platform, Makesense.ai, enabling accurate demarcation of fire boundaries. This resource equips researchers with the necessary data infrastructure to develop innovative methodologies for early fire detection and continuous monitoring, enhancing efforts to protect ecosystems and human lives while promoting sustainable forest management practices. Additionally, the UAVs-FFDB dataset serves as a foundational cornerstone for the advancement and refinement of state-of-the-art AI-based methodologies, aiming to automate fire classification, recognition, detection, and segmentation tasks with unparalleled precision and efficacy. © 2024 The Author(s)