MetaQAP - A Meta-Learning Approach for Quality-Aware Pretraining in Image Quality Assessment
Nisar Ahmed, Gulshan Saleem, Nazik Alturki, Nada Alasbali
TL;DR
This work introduces MetaQAP, a no-reference image quality assessment framework designed to generalize across authentic distortions. It combines quality-aware pre-training on a distortion-rich synthetic dataset, a quality-aware loss that blends $MSE$ with a differentiable surrogate of $SROCC$, and a meta-learning ensemble (via Stepwise Linear Regression) to fuse multiple base models into a final perceptual quality score. Across LiveCD, KonIQ-10K, and BIQ2021, MetaQAP achieves superior $PLCC$ and $SROCC$ scores and demonstrates strong cross-dataset generalization, with ablations confirming the critical roles of the QA loss, pre-training, and meta-learner. The framework advances practical NR-IQA by addressing authentic distortions and offering a robust, efficient ensemble strategy suitable for real-world quality assessment tasks.
Abstract
Image Quality Assessment (IQA) is a critical task in a wide range of applications but remains challenging due to the subjective nature of human perception and the complexity of real-world image distortions. This study proposes MetaQAP, a novel no-reference IQA model designed to address these challenges by leveraging quality-aware pre-training and meta-learning. The model performs three key contributions: pre-training Convolutional Neural Networks (CNNs) on a quality-aware dataset, implementing a quality-aware loss function to optimize predictions, and integrating a meta-learner to form an ensemble model that effectively combines predictions from multiple base models. Experimental evaluations were conducted on three benchmark datasets: LiveCD, KonIQ-10K, and BIQ2021. The proposed MetaQAP model achieved exceptional performance with Pearson Linear Correlation Coefficient (PLCC) and Spearman Rank Order Correlation Coefficient (SROCC) scores of 0.9885/0.9812 on LiveCD, 0.9702/0.9658 on KonIQ-10K, and 0.884/0.8765 on BIQ2021, outperforming existing IQA methods. Cross-dataset evaluations further demonstrated the generalizability of the model, with PLCC and SROCC scores ranging from 0.6721 to 0.8023 and 0.6515 to 0.7805, respectively, across diverse datasets. The ablation study confirmed the significance of each model component, revealing substantial performance degradation when critical elements such as the meta-learner or quality-aware loss function were omitted. MetaQAP not only addresses the complexities of authentic distortions but also establishes a robust and generalizable framework for practical IQA applications. By advancing the state-of-the-art in no-reference IQA, this research provides valuable insights and methodologies for future improvements and extensions in the field.