期刊
IEEE TRANSACTIONS ON MULTIMEDIA
卷 23, 期 -, 页码 4326-4340出版社
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/TMM.2020.3040529
关键词
Task analysis; Distortion; Image quality; Databases; Image recognition; Feature extraction; Deep learning; Transitive transfer learning; image quality assessment; auxiliary domain; distortion translation; semantic feature transfer; generative adversarial network
资金
- National Science Foundation of China [62071369]
This paper proposes a new transitive transfer learning method for no-reference image quality assessment (TTL-IQA) to address the overfitting problem in deep learning-based image quality assessment (IQA) due to limited training samples. The method utilizes multi-domain transitive transfer learning and a generative adversarial network based on distortion translation to construct auxiliary domains and tasks, as well as optimize shared features using a TTL network. Experimental results demonstrate the superiority of the proposed method compared to existing methods and its strong generalization ability.
Image quality assessment (IQA) based on deep learning faces the overfitting problem due to limited training samples available in existing IQA databases. Transfer learning is a plausible solution to the problem, in which the shared features derived from the large-scale Imagenet source domain could be transferred from the original recognition task to the intended IQA task. However, the Imagenet source domain and the IQA target domain as well as their corresponding tasks are not directly related. In this paper, we propose a new transitive transfer learning method for no-reference image quality assessment (TTL-IQA). First, the architecture of the multi-domain transitive transfer learning for IQA is developed to transfer the Imagenet source domain to the auxiliary domain, and then to the IQA target domain. Second, the auxiliary domain and the auxiliary task are constructed by a new generative adversarial network based on distortion translation (DT-GAN). Furthermore, a TTL network of the semantic features transfer (SFTnet) is proposed to optimize the shared features for the TTL-IQA. Experiments are conducted to evaluate the performance of the proposed method on various IQA databases, including the LIVE, TID2013, CSIQ, LIVE multiply distorted and LIVE challenge. The results show that the proposed method significantly outperforms the state-of-the-art methods. In addition, our proposed method demonstrates a strong generalization ability.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据