Semi-Supervised Deep Ensembles for Blind Image Quality Assessment
This work addresses the challenge of creating more generalizable BIQA models for image processing applications, representing an incremental improvement by integrating semi-supervised learning into ensemble methods.
The paper tackled the problem of improving blind image quality assessment (BIQA) by developing a semi-supervised ensemble learning strategy that uses both labeled and unlabeled data to enhance model generalization and failure identification, achieving advantages in these areas as demonstrated through extensive experiments.
Ensemble methods are generally regarded to be better than a single model if the base learners are deemed to be "accurate" and "diverse." Here we investigate a semi-supervised ensemble learning strategy to produce generalizable blind image quality assessment models. We train a multi-head convolutional network for quality prediction by maximizing the accuracy of the ensemble (as well as the base learners) on labeled data, and the disagreement (i.e., diversity) among them on unlabeled data, both implemented by the fidelity loss. We conduct extensive experiments to demonstrate the advantages of employing unlabeled data for BIQA, especially in model generalization and failure identification.