What is the input of the model when training with Adobe5K dataset?

caiyuanhao1998 / Retinexformer

"Retinexformer: One-stage Retinex-based Transformer for Low-light Image Enhancement" (ICCV 2023) & (NTIRE 2024 Challenge)

https://arxiv.org/abs/2303.06705

MIT License

898 stars 78 forks source link

What is the input of the model when training with Adobe5K dataset? #25

Closed kravrolens closed 1 year ago

kravrolens commented 1 year ago

Hi, I would like to know what the input (low-light images) of the model is when training with the Adobe 5K dataset. Or how do you get the input images?

Looking forward to your early reply! THANKS!

caiyuanhao1998 commented 1 year ago

Hi, the input of the model when training with Adobe 5K dataset are the low-light images of 5K. The input images are obtained by exporting the raw images of 5K with the sRGB mode.

Please read this issue for more details

We also provide the exported RGB low-/normal-light image pairs in the Baidu Disk (code:cyh2) and Google Drive for your convenience to download. Please feel free to check and use them.