Bilel Benjdira
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
Bilel Benjdira has a total of 88 co-authors. Here are the top co-authors by number of shared publications:
- Anis Koubaa
Alfaisal University
29 shared publications - Adel Ammar
Prince Sultan University
17 shared publications - Wadii Boulila
Prince Sultan University
14 shared publications - Anas M. Ali
Robotics and Internet-of-Things Laboratory, Prince Sultan University, Riyadh 12435, Saudi Arabia
12 shared publications - Yakoub Bazi
Department of Computer Engineering, College of Computer and Information Sciences, King Saud University, Riyadh, Saudi Arabia
6 shared publications Anis Koubaa
4 shared publications
Latest Publications
- 29 October 2026
- Published by Elsevier in Expert Systems with Applications
- Vol. 334
Leakage-aware calibrated decision-level fusion for mental-health text classification
- 29 October 2026
- Published by Elsevier in Expert Systems with Applications
- Vol. 334
Leakage-aware calibrated decision-level fusion for mental-health text classification
- Open Access
A Novel E-Commerce Search Engine Based on Multimodal Generative AI for Improving Product Accessibility
E-commerce search engines often fall short in delivering accurate and fair results due to their reliance on keyword matching and bias towards larger companies. This paper introduces a novel search engine architecture that integrates advanced AI techniques to address these limitations. Our approach combines Large Language Models (LLMs) for semantic query understanding and Multimodal LLMs (specifically LLaVA) for extracting detailed information from product images to enrich product descriptions. More specifically, we employ a fine-tuned version of the CLIP model to create a unified embedding of product images and descriptions, storing them in a vector database to enable efficient semantic search. Moreover, we fine-tuned the LLaVA model on e-commerce-specific data, significantly improving its performance in generating accurate product descriptions, which were also added to the vector database. The integration of computer vision (CV) techniques allows for a better interpretation of visual product attributes, while the use of vector databases facilitates rapid and relevant search results. Our system also incorporates a recommendation mechanism that leverages these advanced models to provide wider product recommendations. Our experiments demonstrate substantial improvements: the fine-tuned LLaVA model achieved an 80.57% human evaluation rating compared to 62.89% for the original model. Moreover, the overall search accuracy increased after fine-tuning the system by 25 percent points, from 70% to 95%. Our work promotes a more equitable platform for e-commerce search engines while significantly improving user satisfaction.
Open AccessA Novel E-Commerce Search Engine Based on Multimodal Generative AI for Improving Product Accessibility
E-commerce search engines often fall short in delivering accurate and fair results due to their reliance on keyword matching and bias towards larger companies. This paper introduces a novel search engine architecture that integrates advanced AI techniques to address these limitations. Our approach combines Large Language Models (LLMs) for semantic query understanding and Multimodal LLMs (specifically LLaVA) for extracting detailed information from product images to enrich product descriptions. More specifically, we employ a fine-tuned version of the CLIP model to create a unified embedding of product images and descriptions, storing them in a vector database to enable efficient semantic search. Moreover, we fine-tuned the LLaVA model on e-commerce-specific data, significantly improving its performance in generating accurate product descriptions, which were also added to the vector database. The integration of computer vision (CV) techniques allows for a better interpretation of visual product attributes, while the use of vector databases facilitates rapid and relevant search results. Our system also incorporates a recommendation mechanism that leverages these advanced models to provide wider product recommendations. Our experiments demonstrate substantial improvements: the fine-tuned LLaVA model achieved an 80.57% human evaluation rating compared to 62.89% for the original model. Moreover, the overall search accuracy increased after fine-tuning the system by 25 percent points, from 70% to 95%. Our work promotes a more equitable platform for e-commerce search engines while significantly improving user satisfaction.
- 4 July 2026
- Published by Elsevier in Internet of Things
- Vol. 39
Small Language Models for On-Device IoT Intelligence: A Survey on Architectures, Optimization, and Deployment
- 4 July 2026
- Published by Elsevier in Internet of Things
- Vol. 39
Small Language Models for On-Device IoT Intelligence: A Survey on Architectures, Optimization, and Deployment
- Open Access
- 22 May 2026
- Published by Elsevier in Results in Engineering
- Vol. 30
Guided frequency loss for multimedia image enhancement and restoration
- 22 May 2026
- Published by Elsevier in Results in Engineering
- Vol. 30
Open AccessGuided frequency loss for multimedia image enhancement and restoration
- Open Access
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- p. 1-16
MAG-ViT: Multi-Attention Grid Vision Transformer for Efficient and Classification-Aware Remote Sensing Super Resolution
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
- p. 1-16
Open AccessMAG-ViT: Multi-Attention Grid Vision Transformer for Efficient and Classification-Aware Remote Sensing Super Resolution
- Open Access
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Open Journal of Intelligent Transportation Systems
- Vol. 7, 2045-2058
HARNES: Heuristic Algorithm for Routing and Network Efficiency Using SDN in Intelligent Transportation Systems
- 1 January 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Open Journal of Intelligent Transportation Systems
- Vol. 7, 2045-2058
Open AccessHARNES: Heuristic Algorithm for Routing and Network Efficiency Using SDN in Intelligent Transportation Systems
- 31 December 2025
- Published by Elsevier in Expert Systems with Applications
- Vol. 307
SSVA: Self-scanned visual attention for enhanced mask-free shadow removal
- 31 December 2025
- Published by Elsevier in Expert Systems with Applications
- Vol. 307
SSVA: Self-scanned visual attention for enhanced mask-free shadow removal
- 27 August 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. 63 (01962892) , 1-15
Dehazing-DiffGAN: Sequential Fusion of Diffusion Models and GANs for High-Fidelity Remote Sensing Image Dehazing
Atmospheric haze significantly reduces image quality by obscuring important visual details, which in turn impairs the performance of computer vision tasks. Many existing dehazing methods struggle in heavy haze conditions, often resulting in images with poor visual quality, unrealistic features, or visible artifacts. To overcome these limitations, we propose Dehazing-DiffGAN, a novel deep learning framework that combines diffusion models with generative adversarial networks (GANs) in a dual-phase pipeline. In the Dehazing Phase, the diffusion model progressively removes haze by simulating atmospheric scattering, effectively capturing both local structures and global context. In the Enhancement Phase, a GAN-based enhancement module improves image quality by emphasizing fine details and textures, enabling precise restoration of edges and high-frequency components. This synergistic fusion approach allows Dehazing-DiffGAN to generate images of superior visual quality, often indistinguishable from real, haze-free images. Extensive experiments show that Dehazing-DiffGAN achieves state-of-the-art performance on several challenging benchmarks. On the high-resolution SateHaze1k dataset for remote sensing, our model achieves a peak signal-to-noise ratio (PSNR) of 27.21 dB, outperforming the next best method at 26.84 dB, and obtains a superior Learned Perceptual Image Patch Similarity (LPIPS) score of 0.0472 compared to 0.08. On the ultra-high-resolution DNH-HAZE dataset (6000×4000 pixels), Dehazing-DiffGAN achieves the best LPIPS score of 0.267, significantly improving over the second-best score of 0.326. These results highlight the effectiveness of Dehazing-DiffGAN in improving image clarity and confirm its robustness and adaptability across a variety of dehazing tasks. The code is available at https://github.com/riotulab/Dehazing-DiffGAN.
- 27 August 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Geoscience and Remote Sensing
- Vol. 63 (01962892) , 1-15
Dehazing-DiffGAN: Sequential Fusion of Diffusion Models and GANs for High-Fidelity Remote Sensing Image Dehazing
Atmospheric haze significantly reduces image quality by obscuring important visual details, which in turn impairs the performance of computer vision tasks. Many existing dehazing methods struggle in heavy haze conditions, often resulting in images with poor visual quality, unrealistic features, or visible artifacts. To overcome these limitations, we propose Dehazing-DiffGAN, a novel deep learning framework that combines diffusion models with generative adversarial networks (GANs) in a dual-phase pipeline. In the Dehazing Phase, the diffusion model progressively removes haze by simulating atmospheric scattering, effectively capturing both local structures and global context. In the Enhancement Phase, a GAN-based enhancement module improves image quality by emphasizing fine details and textures, enabling precise restoration of edges and high-frequency components. This synergistic fusion approach allows Dehazing-DiffGAN to generate images of superior visual quality, often indistinguishable from real, haze-free images. Extensive experiments show that Dehazing-DiffGAN achieves state-of-the-art performance on several challenging benchmarks. On the high-resolution SateHaze1k dataset for remote sensing, our model achieves a peak signal-to-noise ratio (PSNR) of 27.21 dB, outperforming the next best method at 26.84 dB, and obtains a superior Learned Perceptual Image Patch Similarity (LPIPS) score of 0.0472 compared to 0.08. On the ultra-high-resolution DNH-HAZE dataset (6000×4000 pixels), Dehazing-DiffGAN achieves the best LPIPS score of 0.267, significantly improving over the second-best score of 0.326. These results highlight the effectiveness of Dehazing-DiffGAN in improving image clarity and confirm its robustness and adaptability across a variety of dehazing tasks. The code is available at https://github.com/riotulab/Dehazing-DiffGAN.
- 22 August 2025
- Published by Springer Nature
- p. 129-177
Diffusion Models in a Nutshell: A Tutorial
- 22 August 2025
- Published by Springer Nature
- p. 107-127
Dual-Stage Global and Local Feature Framework for Image Dehazing
- 22 August 2025
- Published by Springer Nature
- p. 107-127
Dual-Stage Global and Local Feature Framework for Image Dehazing