Hi,
According to the descriptions in the Track2 - 'Target domain', the ground truth images of val dataset will be released when the test phase starts. So, I am confused that whether the ground truth images will be provided. And if these gt images are not provided, how do we quantitatively eval our methods except the perceptual quality ?
Looking forward to your reply.
Hi. Instead of releasing the validation ground truth, we have reopened to validation server, so you can still submit your results there and get the PSNR/SSIM scores. Note however that the final evaluation will be performed in terms of the perceptual quality.
Sorry for the incorrect information about the release of the validation ground truth on the codalab page, it should have been corrected now.
Posted by: mdanelljan @ Sept. 5, 2019, 12:30 p.m.