Continual Learning for Blind Image Quality Assessment

Research output: Journal Publications and ReviewsRGC 21 - Publication in refereed journalpeer-review

51 Scopus Citations
View graph of relations

Author(s)

  • Weixia Zhang
  • Dingquan Li
  • Chao Ma
  • Guangtao Zhai
  • Xiaokang Yang

Related Research Unit(s)

Detail(s)

Original languageEnglish
Pages (from-to)2864-2878
Journal / PublicationIEEE Transactions on Pattern Analysis and Machine Intelligence
Volume45
Issue number3
Online published30 May 2022
Publication statusPublished - Mar 2023

Abstract

The explosive growth of image data facilitates the fast development of image processing and computer vision methods for emerging visual applications, meanwhile introducing novel distortions to the processed images. This poses a grand challenge to existing blind image quality assessment (BIQA) models, which are weak at adapting to subpopulation shift. In this paper, we formulate continual learning for BIQA, where a model learns continually from a stream of IQA datasets, building on what was learned from previously seen data. We first identify five desiderata in the continual setting with three criteria to quantify the prediction accuracy, the plasticity, and the stability, respectively. We then propose a simple yet effective continual learning method for BIQA. Specifically, based on a shared backbone network, we add a prediction head for a new dataset, and enforce a regularizer to allow all prediction heads to evolve with new data while being resistant to catastrophic forgetting of old data. We compute the overall quality score by a weighted summation of predictions from all heads. Extensive experiments demonstrate the promise of the proposed continual learning method in comparison to standard training techniques for BIQA, with and without experience replay. We made the code publicly available at https://github.com/zwx8981/BIQA_CL.

Research Area(s)

  • Biological system modeling, Blind image quality assessment, Computational modeling, continual learning, Databases, Distortion, Image quality, Robustness, subpopulation shift, Training

Citation Format(s)

Continual Learning for Blind Image Quality Assessment. / Zhang, Weixia; Li, Dingquan; Ma, Chao et al.
In: IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 45, No. 3, 03.2023, p. 2864-2878.

Research output: Journal Publications and ReviewsRGC 21 - Publication in refereed journalpeer-review