search

khuhm / W-Net

W-Net: Two-Stage U-Net With Misaligned Data for Raw-to-RGB Mapping
MIT License
6 stars 3 forks source link

RAW files passed in test.py #2

Open acehu970601 opened 4 years ago

acehu970601 commented 4 years ago

Hi, thanks for sharing! I am wondering what raw files you passed in test.py, the raw data I used are png files in the original dataset with size 1x448x448. Since the network need 4 channels input, I transform the original file into 4x224x224 RGGB channels, but results I got are confusing, they seem not right. Therefore, I am wondering what files you used for test.py. Thanks a lot!

khuhm commented 4 years ago

Hi, I used the data provided in the "AIM 2019 RAW to RGB Mapping Challenge". In this challenge, the input raw images of size 224x224x4 with Bayer channels [R, Gr, B, Gb] in png format were provided. Hope this helps.

Python-program-maker commented 3 years ago

Hi, I have RAW images from a DSLR camera Canon (image.CR2). How can I feed my image to the test_fullres? I need to transform it in .png (and I already did), but since they are 6000X4000 px colour size, i see the error: "RuntimeError: The size of tensor a (3) must match the size of tensor b (4) at non-singleton dimension 0"

Any tips to resolve it?

khuhm commented 3 years ago

Hi, input raw images should be 4-channel Bayer images ([R, Gr, B, Gb]), not the 3-channel.