search

leonmakise / ELMformer

ELMformer: Efficient Raw Image Restoration with a Locally Multiplicative Transformer
17 stars 0 forks source link

How to pack SIDD benchmark set? #1

Open madfff opened 1 year ago

madfff commented 1 year ago

SIDD dataset does not provide CFA pattern. I am not sure how ot split the one channel raw image into four channels.

ps: looking forward to your code.

leonmakise commented 1 year ago

Hi madfff, the CFA pattern may be useful for demosaic, but our work is to make raw images better whose input and output are one channel. So you do not need to define what color a specific pixel is, and you can simply pack it according to a 2*2 pattern.

A ref from SID: https://github.com/cchen156/Learning-to-See-in-the-Dark/blob/607d8d6b9321a6af28312167e3793bc5b94b7742/train_Sony.py#L85

Miner-AI commented 1 year ago

can you release your code? thank you

junnjxx commented 1 year ago

can you release your code? thank you