search

SSL92 / hyperIQA

Source code for the CVPR'20 paper "Blindly Assess Image Quality in the Wild Guided by A Self-Adaptive Hyper Network"
MIT License
346 stars 52 forks source link

Problems of Implementation Details. Some difference? #4

Closed lllllllllllll-llll closed 3 years ago

lllllllllllll-llll commented 3 years ago

The data augmentation part of this code is different from the implementation details. It says: "we randomly sample and horizontally flipping 25 patches with size 224x224 pixels ... ", and "During testing stage, 25 patches with 224x224 pixels from test image are randomly sampled and their corresponding ... ", but I can not find the part about randomly crop 25 patches, can you point out where this part is? What I find just the normal torchvision.transforms.Compose from data_loader.py instead of randomly crop 25 patches.

SSL92 commented 3 years ago

The data augmentation part of this code is different from the implementation details. It says: "we randomly sample and horizontally flipping 25 patches with size 224x224 pixels ... ", and "During testing stage, 25 patches with 224x224 pixels from test image are randomly sampled and their corresponding ... ", but I can not find the part about randomly crop 25 patches, can you point out where this part is? What I find just the normal torchvision.transforms.Compose from data_loader.py instead of randomly crop 25 patches.

Data augmentation is conducted in folders.py, see line 50, 98 for LIVE and LIVEC data, for instance. You can also alter this setting by changing the '--train_patch_num' and '--test_patch_num' options.

lllllllllllll-llll commented 3 years ago

roger that