Abderrezzaq Sendjasni

Abderrezzaq Sendjasni, Seif-Eddine Benkabou, Mohamed-Chaker Larabi

This article presents a novel approach to improving the accuracy of 360-degree perceptual image quality assessment (IQA) through a two-fold patch selection process. Our methodology combines visual patch selection with embedding similarity-based refinement. The first stage focuses on selecting patches from 360-degree images using three distinct sampling methods to ensure comprehensive coverage of visual content for IQA. The second stage, which is the core of our approach, employs an embedding similarity-based selection process to filter and prioritize the most informative patches based on their embeddings similarity distances. This dual selection mechanism ensures that the training data is both relevant and informative, enhancing the model's learning efficiency. Extensive experiments and statistical analyses using three distance metrics across three benchmark datasets validate the effectiveness of our selection algorithm. The results highlight its potential to deliver robust and accurate 360-degree IQA, with performance gains of up to 4.5% in accuracy and monotonicity of quality score prediction, while using only 40% to 50% of the training patches. These improvements are consistent across various configurations and evaluation metrics, demonstrating the strength of the proposed method. The code for the selection process is available at: https://github.com/sendjasni/patch-selection-360-image-quality.

Abderrezzaq Sendjasni, Mohamed-Chaker Larabi

The rapid evolution of Generative AI (GenAI) models has led to synthetic images of unprecedented realism, challenging traditional methods for distinguishing them from natural photographs. While existing detectors often rely on single-feature spaces, such as statistical regularities, semantic embeddings, or texture patterns, these approaches tend to lack robustness when confronted with diverse and evolving generative models. In this work, we investigate and systematically evaluate a multi-feature fusion framework that combines complementary cues from three distinct spaces: (1) Mean Subtracted Contrast Normalized (MSCN) features capturing low-level statistical deviations; (2) CLIP embeddings encoding high-level semantic coherence; and (3) Multi-scale Local Binary Patterns (MLBP) characterizing mid-level texture anomalies. Through extensive experiments on four benchmark datasets covering a wide range of generative models, we show that individual feature spaces exhibit significant performance variability across different generators. Crucially, the fusion of all three representations yields superior and more consistent performance, particularly in a challenging mixed-model scenario. Compared to state-of-the-art methods, the proposed framework yields consistently improved performance across all evaluated datasets. Overall, this work highlights the importance of hybrid representations for robust GenAI image detection and provides a principled framework for integrating complementary visual cues.

Abderrezzaq Sendjasni, Mohamed-Chaker Larabi

Super-resolution (SR) techniques play a pivotal role in enhancing the quality of low-resolution images, particularly for applications such as security and surveillance, where accurate license plate recognition is crucial. This study proposes a novel framework that combines pixel-based loss with embedding similarity learning to address the unique challenges of license plate super-resolution (LPSR). The introduced pixel and embedding consistency loss (PECL) integrates a Siamese network and applies contrastive loss to force embedding similarities to improve perceptual and structural fidelity. By effectively balancing pixel-wise accuracy with embedding-level consistency, the framework achieves superior alignment of fine-grained features between high-resolution (HR) and super-resolved (SR) license plates. Extensive experiments on the CCPD and PKU dataset validate the efficacy of the proposed framework, demonstrating consistent improvements over state-of-the-art methods in terms of PSNR, SSIM, LPIPS, and optical character recognition (OCR) accuracy. These results highlight the potential of embedding similarity learning to advance both perceptual quality and task-specific performance in extreme super-resolution scenarios.