Support specifying scales in preprocessing div2k dataset (#472)

* First commit!
Add a new scripts to generate training data pairs for blind super resolution

* Second commit!
made some modification

* Third commit
1.remove preprocess_div2k_dataset_bsr.py;
2.add a new argument into preprocess_div2k_dataset.py script to control
whether crop LR images;
3.modify corresponding README.md and README_zh-CN.md.

* Fourth Commit
Replace custom-degradation argument with scale

* Fifth Commit
Restore README.md and README_zh-CN.md

* Sixth Commit
Restore README.md and README_zh-CN.md

* Update README_zh-CN.md

* update annotations

* scale -> scales

Co-authored-by: lizz <lizz@sensetime.com>

tools/data/super-resolution/div2k/preprocess_div2k_dataset.py

@@ -22,6 +22,9 @@ def main_extract_subimages(args):
             A higher value means a smaller size and longer compression time.
             Use 0 for faster CPU decompression. Default: 3, same in cv2.
 
+        scales (list[int]): The downsampling factors corresponding to the
+            LR folders you want to process.
+            Default: [2, 3, 4].
         input_folder (str): Path to the input folder.
         save_folder (str): Path to save folder.
         crop_size (int): Crop size.
@@ -31,14 +34,16 @@ def main_extract_subimages(args):
 
     Usage:
         For each folder, run this script.
-        Typically, there are four folders to be processed for DIV2K dataset.
+        By default, there are four folders to be processed for DIV2K dataset
+            according to scale factor list ([2,3,4])
             DIV2K_train_HR
             DIV2K_train_LR_bicubic/X2
             DIV2K_train_LR_bicubic/X3
             DIV2K_train_LR_bicubic/X4
         After process, each sub_folder should have the same number of
-        subimages.
-        Remember to modify opt configurations according to your settings.
+        subimages. You can also specify scales by modifying the argument
+        'scales'. Remember to modify opt configurations according to your
+        settings.
     """
 
     opt = {}
@@ -53,7 +58,7 @@ def main_extract_subimages(args):
     opt['thresh_size'] = args.thresh_size
     extract_subimages(opt)
 
-    for scale in [2, 3, 4]:
+    for scale in args.scales:
         opt['input_folder'] = osp.join(args.data_root,
                                        f'DIV2K_train_LR_bicubic/X{scale}')
         opt['save_folder'] = osp.join(args.data_root,
@@ -344,6 +349,8 @@ def parse_args():
         description='Prepare DIV2K dataset',
         formatter_class=argparse.ArgumentDefaultsHelpFormatter)
     parser.add_argument('--data-root', help='dataset root')
+    parser.add_argument(
+        '--scales', nargs='*', default=[2, 3, 4], help='scale factor list')
     parser.add_argument(
         '--crop-size',
         nargs='?',