NAFNet

NAFNet copied to clipboard

can you put complete code to train network on our dataset??

Hi, mammad7676,

This repo is based on BasicSR, and it is recommended to refer to the dataset preparation doc and choose 'disk' (which is the simplest) as an example.

The main steps are as follows,

1. prepare your dataset, and make sure the input images and target images are organized by the same image name. e.g.

INPUT_FOLDER/A.png
INPUT_FOLDER/B.png
INPUT_FOLDER/C.png
....

TARGET_FOLDER/A.png
TARGET_FOLDER/B.png
TARGET_FOLDER/C.png
....

2. The modification of the training config.

• the training directories
• io backend, type: disk
• crop size and augmentation
• the validation config. Same as training.

Hope that helps you!