We derive a fundamental property of the posterior distribution in Gaussian denoising, and use it to propose a new way for uncertainty visualization, which requires no training or fine-tuning.
uncertainty uncertainty-quantification uncertainty-estimation denoising-images denoising posterior-distributions denoising-network uncertainty-visualisation semantic-uncertainty
-
Updated
Apr 30, 2024 - Jupyter Notebook