What is it about?
Over the past 5 years, several methods have attempted to solve the SI-HDR problem and each successive method claims to improve the state-of-the-art by showing better automatic quality metric scores. We compare the methods in a subjective experiment and improve the reliability of these metrics.
Featured Image
Photo by Emre Karataş on Unsplash
Why is it important?
Our findings show that human observers do not really agree with widely-used metrics for evaluating SI-HDR methods. We introduce a new evaluation dataset and protocol to benchmark future works.
Read the Original
This page is a summary of: Comparison of single image HDR reconstruction methods — the caveats of quality assessment, August 2022, ACM (Association for Computing Machinery),
DOI: 10.1145/3528233.3530729.
You can read the full text:
Resources
Contributors
The following have contributed to this page