search

megvii-model / HINet

Other
367 stars 62 forks source link

About datasets #28

Open jiaaihhy opened 3 years ago

jiaaihhy commented 3 years ago

dear author: I have a question that training on gopro, have 2 blur folder, that blur and blur_gamma, I want to know you used is blur or blur_gamma? thankyou

mayorx commented 3 years ago

Hi, jiaaihhy, Thanks for your attention to HINet. For a fair comparison, we simply adopt the GoPro dataset provided by MPRNet.

leefly072 commented 3 years ago

Dear author: I have a question that training on SIDD, have 2 val/gt_crop.lmdb and val/input_crops.lmdb folder. how can I get it, can you tell me? Thank you!

mayorx commented 3 years ago

Dear author: I have a question that training on SIDD, have 2 val/gt_crop.lmdb and val/input_crops.lmdb folder. how can I get it, can you tell me? Thank you!

Hi, leefly072,

In https://github.com/megvii-model/HINet#image-restoration-tasks, (Image Denoise - SIDD dataset), when you prepared the SIDD dataset, simply run python scripts/data_preparation/sidd.py , it would crop the images into patches and make the lmdb data.