English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Learning based image quality assessment (IQA) models have obtained impressiveperformance with the help of reliable subjective quality labels, where meanopinion score (MOS) is the most popular choice. However, in view of thesubjective bias of individual annotators, the labor-abundant MOS (LA-MOS)typically requires a large collection of opinion scores from multipleannotators for each image, which significantly increases the learning cost. Inthis paper, we aim to learn robust IQA models from low-cost MOS (LC-MOS), whichonly requires very few opinion scores or even a single opinion score for eachimage. More specifically, we consider the LC-MOS as the noisy observation ofLA-MOS and enforce the IQA model learned from LC-MOS to approach the unbiasedestimation of LA-MOS. In this way, we represent the subjective bias betweenLC-MOS and LA-MOS, and the model bias between IQA predictions learned fromLC-MOS and LA-MOS (i.e., dual-bias) as two latent variables with unknownparameters. By means of the expectation-maximization based alternatingoptimization, we can jointly estimate the parameters of the dual-bias, whichsuppresses the misleading of LC-MOS via a gated dual-bias calibration (GDBC)module. To the best of our knowledge, this is the first exploration of robustIQA model learning from noisy low-cost labels. Theoretical analysis andextensive experiments on four popular IQA datasets show that the proposedmethod is robust toward different bias rates and annotation numbers andsignificantly outperforms the other learning based IQA models when only LC-MOSis available. Furthermore, we also achieve comparable performance with respectto the other models learned with LA-MOS.

Learning with Noisy Low-Cost MOS for Image Quality Assessment via Dual-Bias Calibration