search

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
925 stars 82 forks source link

Data set problem #107

Closed WALYP closed 2 months ago

WALYP commented 2 months ago

Besides comparing the PSNR of strong and weak photos, can this code handle its own images? Or can you input a data set and get an enhanced data set? Only enhance the picture.

caiyuanhao1998 commented 2 months ago

Hi, thanks for your interest,

Yes, the first step is to pick up a pre-trained weight. Then use the corresponding config file. Modify the input and gt paths in the config file to the path of your data set. Finally run the test_from_dataset.py.

If you find our repo useful, please help us star it. Thank you :)

WALYP commented 2 months ago

Thank you very much for your reply! I still have some doubts. The dataroot_gt and dataroot_lq paths in the configuration file are high-quality and low-quality pictures respectively. We can compare them with the images in the dataroot_gt folder by enhancing the images in the dataroot_lq folder. Through test_from_dataset.py, we can get data such as psnr, and in the result, we can get the different images of the contrast image, but what I want to ask is how can I get the enhanced image. It is an image compared with the dataroot_gt folder.Thank you for your answer!