Metrics Range and value interpretation
pentanol2 opened this issue · comments
YOUSSEF AIDANI commented
I have a suggestion here.
Can you add the value range of each metric and if the higher is better or if the opposite is true ?
https://github.com/chaofengc/Awesome-Image-Quality-Assessment#no-reference-nr
Chaofeng Chen commented
Thanks for suggestion.
It is not possible to give range or better value indications at this repo, especially for recent DL methods. Because for learning based approaches, it depends on the dataset labeling and how they train their network.
Therefore, we must have the trained model first to give the right answer to your question. You may refer to our IQA-PyTorch toolbox. We have a lower_better
attribute for each implemented metric.