search

zwx8981 / UNIQUE

The repository for 'Uncertainty-aware blind image quality assessment in the laboratory and wild' and 'Learning to blindly assess image quality in the laboratory and wild'
Apache License 2.0
131 stars 11 forks source link

How are the cross-dataset experiments conducted? #18

Closed yunchengwang closed 2 years ago

yunchengwang commented 2 years ago

Hi,

Thank you so much for the amazing work! I have a question regarding Table IV in the original paper.

In the table, baseline models are trained with some datasets and tested on the other datasets. How are the cross-dataset experiments conducted given that the MOS/DMOS ranges are different across datasets? Do you re-train the FC layers for each dataset during testing? Thanks.

zwx8981 commented 2 years ago

Hi, thanks for your interest.

In cross-dataset experiments, we don't need to re-train FC layers for each dataset during testing. We directly use the models trained on other datasets to test on target datasets. The difference of ranges won't affect the results, you can still compute the SRCC/PLCC between yor model predictions and the gound-truth MOS/DMOS.