In the Foggy Cityscapes dataset, the evaluation methods are: (1) use the data of three foggy levels(0.005,0.01,0.02) together as the training set and verification set; (2) use the data set of Foggy level=0.02 as the training set and verification set,
All of the dataset, including all the foggy levels are used. The used COCO style Foggy Cityscapes dataset annotations are provided by the SFA github repo.
In the Foggy Cityscapes dataset, the evaluation methods are: (1) use the data of three foggy levels(0.005,0.01,0.02) together as the training set and verification set; (2) use the data set of Foggy level=0.02 as the training set and verification set,
Which dataset do you use?