Bilal Hassan
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Bilal Hassan has a total of 64 co-authors. Here are the top co-authors by number of shared publications:
Hamid Jahankhani
3 shared publications- Rao Faizan Ali
School of Computing, University of Kent, Canterbury CT2 7NZ, UK
3 shared publications Yusra Siddiqi
METICS Solutions Ltd, London, U.K.
3 shared publications- Hamid Jahankhani
Northumbria University
2 shared publications - Muhammad Farooq Wasiq
METICS Solutions Ltd., 6A Talisman Close Ilford, London IG3 9JA, UK
2 shared publications - Preeti Patel
London Metropolitan University
2 shared publications
Latest Publications
- Open Access
- 30 October 2026
- Published by Wiley in International Journal of Imaging Systems and Technology
- Vol. 36 (6)
ClinX ‐Net: Multimodal Visual–Language Feature Fusion for Chest X‐Ray Report Generation
Automated chest X‐ray report generation has the potential to improve reporting efficiency and reduce variability in radiological documentation. However, existing methods remain challenged by subtle abnormality description, long‐range textual consistency, and effective integration of clinical information with visual findings. To address these limitations, we propose ClinX‐Net, a multimodal framework for chest X‐ray report generation that combines hierarchical visual representation learning, disease‐aware semantic encoding, and language‐model‐based text generation. The visual branch employs ConvNeXt to extract multilevel chest radiograph features, which are integrated through multiscale fusion to capture global anatomical structures and localized pathological patterns. Disease label representations are incorporated to provide additional semantic guidance, while a BioBERT encoder captures clinical context from radiology‐related text. The resulting visual, disease‐semantic, and clinical representations are fused to condition a T5‐based decoder for report generation. Experiments were conducted on the publicly available IU X‐Ray and MIMIC‐CXR datasets. On IU X‐Ray, ClinX‐Net achieved a BLEU‐4 of 0.199, ROUGE‐L score of 0.454 and a METEOR score of 0.222, while on MIMIC‐CXR, it achieved BLEU‐4, ROUGE‐L, and METEOR scores of 0.122, 0.296, and 0.174, respectively. Additional semantic and clinical evaluations using BERTScore and CheXbert demonstrate improved contextual similarity and preservation of clinically relevant observations in generated reports. Qualitative analysis further shows improved anatomical descriptions and better localization of reported findings. These results indicate that integrating hierarchical visual features, disease‐aware semantic representations, and clinical context improves the quality and clinical relevance of automated chest X‐ray report generation.
- 30 October 2026
- Published by Wiley in International Journal of Imaging Systems and Technology
- Vol. 36 (6)
Open AccessClinX ‐Net: Multimodal Visual–Language Feature Fusion for Chest X‐Ray Report Generation
Automated chest X‐ray report generation has the potential to improve reporting efficiency and reduce variability in radiological documentation. However, existing methods remain challenged by subtle abnormality description, long‐range textual consistency, and effective integration of clinical information with visual findings. To address these limitations, we propose ClinX‐Net, a multimodal framework for chest X‐ray report generation that combines hierarchical visual representation learning, disease‐aware semantic encoding, and language‐model‐based text generation. The visual branch employs ConvNeXt to extract multilevel chest radiograph features, which are integrated through multiscale fusion to capture global anatomical structures and localized pathological patterns. Disease label representations are incorporated to provide additional semantic guidance, while a BioBERT encoder captures clinical context from radiology‐related text. The resulting visual, disease‐semantic, and clinical representations are fused to condition a T5‐based decoder for report generation. Experiments were conducted on the publicly available IU X‐Ray and MIMIC‐CXR datasets. On IU X‐Ray, ClinX‐Net achieved a BLEU‐4 of 0.199, ROUGE‐L score of 0.454 and a METEOR score of 0.222, while on MIMIC‐CXR, it achieved BLEU‐4, ROUGE‐L, and METEOR scores of 0.122, 0.296, and 0.174, respectively. Additional semantic and clinical evaluations using BERTScore and CheXbert demonstrate improved contextual similarity and preservation of clinically relevant observations in generated reports. Qualitative analysis further shows improved anatomical descriptions and better localization of reported findings. These results indicate that integrating hierarchical visual features, disease‐aware semantic representations, and clinical context improves the quality and clinical relevance of automated chest X‐ray report generation.
- Open Access
- 25 June 2026
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 21 (6) , e0352091
The role of community capacity building (CCB) in promoting local development: Analyzing leadership in the Hyderabad cantonment community
With leadership positioned as a key tenet, this study investigates the function of Community Capacity Building (CCB) in fostering local development. It combines quantitative survey data from Hyderabad Cantonment, Pakistan, with qualitative interview data using a mixed-methods design. The results show notable deficiencies in women’s leadership participation, resource mobilization, and participatory governance. The study suggests specific tactics to improve leadership for sustainable and inclusive development. The study creates an empirically supported, situation-specific, yet broadly applicable model for grassroots development. Creating localized policy recommendations that address governance deficiencies and integrating leadership roles using both qualitative narratives and quantitative insights are two important contributions. For academics and decision-makers looking to promote participatory governance in urban and semi-urban settings, the results offer useful evidence. It is important to note that findings reflect community perceptions of leadership effectiveness rather than objective performance indicators, and therefore should be interpreted as perception-based evidence rather than causal measurement.
- 25 June 2026
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 21 (6) , e0352091
Open AccessThe role of community capacity building (CCB) in promoting local development: Analyzing leadership in the Hyderabad cantonment community
With leadership positioned as a key tenet, this study investigates the function of Community Capacity Building (CCB) in fostering local development. It combines quantitative survey data from Hyderabad Cantonment, Pakistan, with qualitative interview data using a mixed-methods design. The results show notable deficiencies in women’s leadership participation, resource mobilization, and participatory governance. The study suggests specific tactics to improve leadership for sustainable and inclusive development. The study creates an empirically supported, situation-specific, yet broadly applicable model for grassroots development. Creating localized policy recommendations that address governance deficiencies and integrating leadership roles using both qualitative narratives and quantitative insights are two important contributions. For academics and decision-makers looking to promote participatory governance in urban and semi-urban settings, the results offer useful evidence. It is important to note that findings reflect community perceptions of leadership effectiveness rather than objective performance indicators, and therefore should be interpreted as perception-based evidence rather than causal measurement.
- Open Access
- 19 March 2026
- Published by Springer Nature in Multimedia Tools and Applications
- Vol. 85 (4)
Global soft biometrics in surveillance: benchmark in the field & open challenges
Global soft biometrics-based identification refers to the recognition of subjects using human traits such as gender, age, and ethnicity. Unlike traditional biometric methods that rely on unique physical markers such as fingerprints or iris scans, soft biometrics represents a non-intrusive, viable, and versatile approach, thus making them particularly valuable for surveillance and security applications. Despite significant advances, several issues have been associated with traditional biometrics, like maintaining accuracy, addressing algorithmic bias, and limited computational efficiency. To address those issues, this paper presents a comprehensive coverage of the current advances in Global soft biometric-based recognition as a solution, where four key contributions are made; i.e., (i) advocacy on the relevance and impact of soft biometrics in surveillance and security, (ii) development of a new and unique CeleBImg dataset to overcome algorithmic biases and improve diversity in soft biometric-based recognition, (iii) rigorous performance comparison of current methods in-practice for Global soft biometrics-based recognition and, (iv) identification of open challenges with potential solutions in the field within the context of surveillance and security. This paper sets a solid foundation for using Global soft biometrics in the CCTV-based surveillance and security domain, with their significance, relevance and effectiveness.
- 19 March 2026
- Published by Springer Nature in Multimedia Tools and Applications
- Vol. 85 (4)
Open AccessGlobal soft biometrics in surveillance: benchmark in the field & open challenges
Global soft biometrics-based identification refers to the recognition of subjects using human traits such as gender, age, and ethnicity. Unlike traditional biometric methods that rely on unique physical markers such as fingerprints or iris scans, soft biometrics represents a non-intrusive, viable, and versatile approach, thus making them particularly valuable for surveillance and security applications. Despite significant advances, several issues have been associated with traditional biometrics, like maintaining accuracy, addressing algorithmic bias, and limited computational efficiency. To address those issues, this paper presents a comprehensive coverage of the current advances in Global soft biometric-based recognition as a solution, where four key contributions are made; i.e., (i) advocacy on the relevance and impact of soft biometrics in surveillance and security, (ii) development of a new and unique CeleBImg dataset to overcome algorithmic biases and improve diversity in soft biometric-based recognition, (iii) rigorous performance comparison of current methods in-practice for Global soft biometrics-based recognition and, (iv) identification of open challenges with potential solutions in the field within the context of surveillance and security. This paper sets a solid foundation for using Global soft biometrics in the CCTV-based surveillance and security domain, with their significance, relevance and effectiveness.
- Open Access
Transfer Learning-Based Ethnicity Recognition Using Arbitrary Images Captured Through Diverse Imaging Sensors
Ethnicity recognition has become increasingly important for a wide range of applications, highlighting the need for accurate and robust predictive models. Despite advances in machine learning, ethnicity classification remains a challenging research problem due to variations in facial features, class imbalance, and generalization issues. This study provides a concise synthesis of prior work to motivate the problem and then introduces a novel experimental framework for ethnicity recognition rather than a survey review. It proposes an improved approach that leverages transfer learning to enhance classification performance. The inclusion of various imaging sensors in the proposed methodology allows for an examination of how these imaging sensors impact the performance of facial recognition systems when a variety of images are captured under a number of real-world conditions, using professional and consumer-grade devices to create a range of conditions; from this dataset, the UTKFace dataset will be used to train and validate our method; an additional balanced dataset of Test Celebrities Faces was also created, representing five different ethnic groups (Black, Asian, White, Indian, and Other); the “Other” classification was specifically excluded for final evaluations to eliminate ambiguity and enhance stability. Rigorous preprocessing of both datasets was performed for optimal extraction of features from the sensors’ acquired images; the performance of several pre-trained CNN (Convolutional Neural Network) models (VGG16, DenseNet169, VGG19, ResNet50, MobileNetV2, InceptionV3 and EfficientNetB4) was used to identify an Ideal Hyperparameter Configuration for Optimal Performance. The resulting experimental results indicate that the VGG19 model achieved an 87% validation accuracy and a Maximum test accuracy of 75% on the Primary Dataset of Celebrity Faces; subsequently, the VGG19 model demonstrated a Range of Per-Class Accuracies, in addition to an overall accuracy of 87% across all five ethnic groups (51–90%+). This work demonstrates that leveraging transfer learning on imaging-sensor-captured images enables robust ethnicity classification with high accuracy and improved training efficiency relative to full model retraining. Furthermore, systematic hyperparameter optimization enhances model generalization and mitigates overfitting. Comparative experiments with recent state-of-the-art methods (2023–2025) further confirm that our optimized VGG19 model achieves competitive performance, reinforcing the effectiveness of the proposed reproducible and fairness-aware evaluation framework.
Open AccessTransfer Learning-Based Ethnicity Recognition Using Arbitrary Images Captured Through Diverse Imaging Sensors
Ethnicity recognition has become increasingly important for a wide range of applications, highlighting the need for accurate and robust predictive models. Despite advances in machine learning, ethnicity classification remains a challenging research problem due to variations in facial features, class imbalance, and generalization issues. This study provides a concise synthesis of prior work to motivate the problem and then introduces a novel experimental framework for ethnicity recognition rather than a survey review. It proposes an improved approach that leverages transfer learning to enhance classification performance. The inclusion of various imaging sensors in the proposed methodology allows for an examination of how these imaging sensors impact the performance of facial recognition systems when a variety of images are captured under a number of real-world conditions, using professional and consumer-grade devices to create a range of conditions; from this dataset, the UTKFace dataset will be used to train and validate our method; an additional balanced dataset of Test Celebrities Faces was also created, representing five different ethnic groups (Black, Asian, White, Indian, and Other); the “Other” classification was specifically excluded for final evaluations to eliminate ambiguity and enhance stability. Rigorous preprocessing of both datasets was performed for optimal extraction of features from the sensors’ acquired images; the performance of several pre-trained CNN (Convolutional Neural Network) models (VGG16, DenseNet169, VGG19, ResNet50, MobileNetV2, InceptionV3 and EfficientNetB4) was used to identify an Ideal Hyperparameter Configuration for Optimal Performance. The resulting experimental results indicate that the VGG19 model achieved an 87% validation accuracy and a Maximum test accuracy of 75% on the Primary Dataset of Celebrity Faces; subsequently, the VGG19 model demonstrated a Range of Per-Class Accuracies, in addition to an overall accuracy of 87% across all five ethnic groups (51–90%+). This work demonstrates that leveraging transfer learning on imaging-sensor-captured images enables robust ethnicity classification with high accuracy and improved training efficiency relative to full model retraining. Furthermore, systematic hyperparameter optimization enhances model generalization and mitigates overfitting. Comparative experiments with recent state-of-the-art methods (2023–2025) further confirm that our optimized VGG19 model achieves competitive performance, reinforcing the effectiveness of the proposed reproducible and fairness-aware evaluation framework.
- Open Access
- 19 September 2025
- Published by MDPI AG in Electronics
- Vol. 14 (18) , 3719
Transfer Learning-Based Distance-Adaptive Global Soft Biometrics Prediction in Surveillance
Soft biometric prediction—including age, gender, and ethnicity—is critical in surveillance applications, yet often suffers from performance degradation as the subject-to-camera distance increases. This study hypothesizes that embedding distance-awareness into the training process can mitigate such degradation and enhance model generalization across varying visual conditions. We propose a distance-adaptive, multi-task deep learning framework built upon EfficientNetB3, augmented with task-specific heads and trained progressively across four distance intervals (4 m to 10 m). A weighted composite loss function is employed to balance classification and regression objectives. The model is evaluated on a hybrid dataset combining the Front-View Gait (FVG) and MMV annotated pedestrian datasets, totaling over 19,000 samples. Experimental results demonstrate that the framework achieves up to 95% gender classification accuracy at 4 m and retains 85% accuracy at 10 m. Ethnicity prediction maintains an accuracy above 65%, while age estimation achieves a mean absolute error (MAE) ranging from 1.1 to 1.5 years. These findings validate the model’s robustness across distances and its superiority over conventional static learning approaches. Despite challenges such as computational overhead and annotation demands, the proposed approach offers a scalable and real-time-capable solution for distance-resilient biometric systems.
- 19 September 2025
- Published by MDPI AG in Electronics
- Vol. 14 (18) , 3719
Open AccessTransfer Learning-Based Distance-Adaptive Global Soft Biometrics Prediction in Surveillance
Soft biometric prediction—including age, gender, and ethnicity—is critical in surveillance applications, yet often suffers from performance degradation as the subject-to-camera distance increases. This study hypothesizes that embedding distance-awareness into the training process can mitigate such degradation and enhance model generalization across varying visual conditions. We propose a distance-adaptive, multi-task deep learning framework built upon EfficientNetB3, augmented with task-specific heads and trained progressively across four distance intervals (4 m to 10 m). A weighted composite loss function is employed to balance classification and regression objectives. The model is evaluated on a hybrid dataset combining the Front-View Gait (FVG) and MMV annotated pedestrian datasets, totaling over 19,000 samples. Experimental results demonstrate that the framework achieves up to 95% gender classification accuracy at 4 m and retains 85% accuracy at 10 m. Ethnicity prediction maintains an accuracy above 65%, while age estimation achieves a mean absolute error (MAE) ranging from 1.1 to 1.5 years. These findings validate the model’s robustness across distances and its superiority over conventional static learning approaches. Despite challenges such as computational overhead and annotation demands, the proposed approach offers a scalable and real-time-capable solution for distance-resilient biometric systems.
- Open Access
Cross-Project Multiclass Classification of EARS-Based Functional Requirements Utilizing Natural Language Processing, Machine Learning, and Deep Learning
Software requirements are primarily classified into functional and non-functional requirements. While research has explored automated multiclass classification of non-functional requirements, functional requirements remain largely unexplored. This study addressed that gap by introducing a comprehensive dataset comprising 9529 functional requirements from 315 diverse projects. The requirements are classified into five categories: ubiquitous, event-driven, state-driven, unwanted behavior, and optional capabilities. Natural Language Processing (NLP), machine learning (ML), and deep learning (DL) techniques are employed to enable automated classification. All software requirements underwent several procedures, including normalization and feature extraction techniques such as TF-IDF. A series of Machine learning (ML) and deep learning (DL) experiments were conducted to classify subcategories of functional requirements. Among the trained models, the convolutional neural network achieved the highest performance, with an accuracy of 93, followed by the long short-term memory network with an accuracy of 92, outperforming traditional decision-tree-based methods. This work offers a foundation for precise requirement classification tools by providing both the dataset and an automated classification approach.
Open AccessCross-Project Multiclass Classification of EARS-Based Functional Requirements Utilizing Natural Language Processing, Machine Learning, and Deep Learning
Software requirements are primarily classified into functional and non-functional requirements. While research has explored automated multiclass classification of non-functional requirements, functional requirements remain largely unexplored. This study addressed that gap by introducing a comprehensive dataset comprising 9529 functional requirements from 315 diverse projects. The requirements are classified into five categories: ubiquitous, event-driven, state-driven, unwanted behavior, and optional capabilities. Natural Language Processing (NLP), machine learning (ML), and deep learning (DL) techniques are employed to enable automated classification. All software requirements underwent several procedures, including normalization and feature extraction techniques such as TF-IDF. A series of Machine learning (ML) and deep learning (DL) experiments were conducted to classify subcategories of functional requirements. Among the trained models, the convolutional neural network achieved the highest performance, with an accuracy of 93, followed by the long short-term memory network with an accuracy of 92, outperforming traditional decision-tree-based methods. This work offers a foundation for precise requirement classification tools by providing both the dataset and an automated classification approach.
- Open Access
- 6 June 2025
- Published by MDPI AG in Electronics
- Vol. 14 (12) , 2320
Transformer-Based Abstractive Summarization of Legal Texts in Low-Resource Languages
The emergence of large language models (LLMs) has revolutionized the trajectory of NLP research. Transformers, combined with attention mechanisms, have increased computational power, and massive datasets have led to the emergence of pre-trained large language models (PLLMs), which offer promising possibilities for multilingual applications in low-resource settings. However, the scarcity of annotated resources and suitably pre-trained models continues to pose a significant hurdle for the low-resource abstractive text summarization of legal texts, particularly in Urdu. This study presents a transfer learning approach using pre-trained multilingual large models (the mBART and mT5, Small, Base, and Large) to generate abstractive summaries of Urdu legal texts. A curated dataset was developed with legal experts, who produced ground-truth summaries. The models were fine-tuned on this domain-specific corpus to adapt them for low-resource legal summarization. The experimental results demonstrated that the mT5-Large, fine-tuned on Urdu legal texts, outperforms all other evaluated models across standard summarization metrics, achieving ROUGE-1 scores of 0.7889, ROUGE-2 scores of 0.5961, and ROUGE-L scores of 0.7813. This indicates its strong capacity to generate fluent, coherent, and legally accurate summaries. The mT5-Base model closely follows with ROUGE-1 = 0.7774, while the mT5-Small shows moderate performance (ROUGE-1 = 0.6406), with reduced fidelity in capturing legal structure. The mBART50 model, despite being fine-tuned on the same legal corpus, performs lower (ROUGE-1 = 0.5914), revealing its relative limitations in this domain. Notably, models trained or fine-tuned on non-legal, out-of-domain data, such as the urT5 (ROUGE-1 = 0.3912), the mT5-XLSUM (ROUGE-1 = 0.0582), and the mBART50 (XLSUM) (ROUGE-1 = 0.0545), exhibit poor generalization to legal summaries, underscoring the necessity of domain adaptation when working in low-resource legal contexts. These findings highlight the effectiveness of fine-tuning multilingual LLMs for domain-specific tasks. The gains in legal summarization demonstrate the practical value of transfer learning in low-resource settings and the broader potential of AI-driven tools for legal document processing, information retrieval, and decision support.
- 6 June 2025
- Published by MDPI AG in Electronics
- Vol. 14 (12) , 2320
Open AccessTransformer-Based Abstractive Summarization of Legal Texts in Low-Resource Languages
The emergence of large language models (LLMs) has revolutionized the trajectory of NLP research. Transformers, combined with attention mechanisms, have increased computational power, and massive datasets have led to the emergence of pre-trained large language models (PLLMs), which offer promising possibilities for multilingual applications in low-resource settings. However, the scarcity of annotated resources and suitably pre-trained models continues to pose a significant hurdle for the low-resource abstractive text summarization of legal texts, particularly in Urdu. This study presents a transfer learning approach using pre-trained multilingual large models (the mBART and mT5, Small, Base, and Large) to generate abstractive summaries of Urdu legal texts. A curated dataset was developed with legal experts, who produced ground-truth summaries. The models were fine-tuned on this domain-specific corpus to adapt them for low-resource legal summarization. The experimental results demonstrated that the mT5-Large, fine-tuned on Urdu legal texts, outperforms all other evaluated models across standard summarization metrics, achieving ROUGE-1 scores of 0.7889, ROUGE-2 scores of 0.5961, and ROUGE-L scores of 0.7813. This indicates its strong capacity to generate fluent, coherent, and legally accurate summaries. The mT5-Base model closely follows with ROUGE-1 = 0.7774, while the mT5-Small shows moderate performance (ROUGE-1 = 0.6406), with reduced fidelity in capturing legal structure. The mBART50 model, despite being fine-tuned on the same legal corpus, performs lower (ROUGE-1 = 0.5914), revealing its relative limitations in this domain. Notably, models trained or fine-tuned on non-legal, out-of-domain data, such as the urT5 (ROUGE-1 = 0.3912), the mT5-XLSUM (ROUGE-1 = 0.0582), and the mBART50 (XLSUM) (ROUGE-1 = 0.0545), exhibit poor generalization to legal summaries, underscoring the necessity of domain adaptation when working in low-resource legal contexts. These findings highlight the effectiveness of fine-tuning multilingual LLMs for domain-specific tasks. The gains in legal summarization demonstrate the practical value of transfer learning in low-resource settings and the broader potential of AI-driven tools for legal document processing, information retrieval, and decision support.
- 26 March 2025
- Published by Springer Nature
- p. 605-615
From Distraction to Interaction: Investigating Learner Engagement Challenges in Virtual Classrooms
- 26 March 2025
- Published by Springer Nature
- p. 605-615
From Distraction to Interaction: Investigating Learner Engagement Challenges in Virtual Classrooms