1 code implementation • 5 Aug 2022 • Xigran Liao, Baoliang Chen, Hanwei Zhu, Shiqi Wang, Mingliang Zhou, Sam Kwong
Existing deep learning-based full-reference IQA (FR-IQA) models usually predict the image quality in a deterministic way by explicitly comparing the features, gauging how severely distorted an image is by how far the corresponding feature lies from the space of the reference images.
