Add check_image_shapes script

- this script is useful to check if the dataset contains different sizes of images

script/check_image_shapes.py

@@ -0,0 +1,72 @@
+#!/usr/bin/env python
+import argparse
+from pathlib import Path
+
+import skimage.io
+from tqdm import tqdm
+
+def main():
+    parser = argparse.ArgumentParser()
+    parser.add_argument('dir', help='root path of CTC dataset')
+    parser.add_argument(
+        '--3d',
+        dest='is_3d',
+        action='store_true',
+        help=('specify if generate 3D labels (only used for 3D+time datasets)')
+    )
+    parser.add_argument(
+        '--gt',
+        dest='gt',
+        action='store_true',
+        help=('specify if process GT')
+    )
+    parser.add_argument(
+        '--st',
+        dest='st',
+        action='store_true',
+        help=('specify if process ST')
+    )
+    parser.add_argument(
+        '--gtst',
+        dest='gtst',
+        action='store_true',
+        help=('specify if process GT+ST')
+    )
+    args = parser.parse_args()
+    ref_types_list = []
+    if args.gt:
+        ref_types_list.append(('GT',))
+    if args.st:
+        ref_types_list.append(('ST',))
+    if args.gtst:
+        ref_types_list.append(('GT', 'ST'))
+    if len(ref_types_list) == 0:
+        ref_types_list = [('GT',), ('ST',), ('GT', 'ST')]
+    p = Path(args.dir)
+    is_3d = args.is_3d
+    seg_dir_name = f'SEG{"_3d" if is_3d else ""}'
+    for i in range(2):
+        p_img = p / f'0{i+1}'
+        shape = None
+        dtype = None
+        for f in tqdm(sorted(p_img.glob('*.tif'))):
+            img = skimage.io.imread(str(f))
+            if shape is None:
+                shape = img.shape
+                dtype = img.dtype
+            else:
+                if shape != img.shape:
+                    print(f'image {f.name}: shape expected {shape} but got {img.shape}.')
+                if dtype != img.dtype:
+                    print(f'image {f.name}: dtype expected {dtype} but got {img.dtype}.')
+        for ref_types in ref_types_list:
+            for ref_type in ref_types:
+                p_seg = p / f'0{i+1}_{ref_type}' / seg_dir_name
+                if p_seg.exists():
+                    for f in tqdm(sorted(p_seg.glob('*.tif'))):
+                        img = skimage.io.imread(str(f))
+                        if shape != img.shape:
+                            print(f'image {f.name}: shape expected {shape} but got {img.shape}.')
+
+if __name__ == "__main__":
+    main()