DAEFR_test_datasets / README.md
LIAGM's picture
Update testing dataset descriptions
b0f48ec
|
raw
history blame
1.04 kB
metadata
license: apache-2.0

We evaluate DAEFR on one synthetic dataset CelebA-Test, and two real-world datasets LFW-Test and WIDER-Test.

Datasets Filename Short Description Source
CelebA-Test (HQ) celeba_512_validation.zip 3000 (HQ) ground truth images for evaluation RestoreFormer
CelebA-Test (LQ) self_celeba_512_v2.zip 3000 (LQ) synthetic images for testing Ourselves
LFW-Test (LQ) lfw_cropped_faces.zip 1711 real-world images for testing VQFR
WIDER-Test (LQ) Wider-Test.zip 970 real-world images for testing CodeFormer