improve simple image loader

utils/__init__.py

@@ -4,4 +4,4 @@
 # ================================================================
 
 from iseg.utils.sugars import *
-from iseg.utils.common import resize_image, simple_load_image
+from iseg.utils.common import resize_image

utils/common.py

@@ -73,25 +73,4 @@ def down_size_image_by_scale(images, scale, method=None, name=None):
     target_height = image_height // scale + is_height_odd
     target_width = image_width // scale + is_width_odd
 
-    return resize_image(images=images, size=(target_height, target_width), method=method, name=name)
-
-
-def simple_load_image(image_path, min_size_unit=1):
-
-    image_tensor = tf.image.decode_jpeg(tf.io.read_file(image_path), channels=3)
-    image_tensor = tf.expand_dims(tf.cast(image_tensor, tf.float32), axis=0)  # [1, H, W, 3]
-
-    image_size = tf.shape(image_tensor)[1:3]
-
-    pad_height = tf.cast(tf.math.ceil(image_size[0] / 32) * 32, tf.int32)
-    pad_width = tf.cast(tf.math.ceil(image_size[1] / 32) * 32, tf.int32)
-
-    pad_height = pad_height if pad_height % 2 != 0 else pad_height + 1
-    pad_width = pad_height if pad_width % 2 != 0 else pad_width + 1
-
-    from iseg.data_process.utils import pad_to_bounding_box, normalize_value_range
-
-    pad_image_tensor = pad_to_bounding_box(image_tensor, 0, 0, pad_height, pad_width, pad_value=[127.5, 127.5, 127.5])
-    pad_image_tensor = normalize_value_range(pad_image_tensor)
-
-    return pad_image_tensor, image_size
+    return resize_image(images=images, size=(target_height, target_width), method=method, name=name)

utils/data_loader.py

@@ -2,6 +2,8 @@
 
 from PIL import Image
 
+from iseg.data_process.augments.pad_to_odd_augment import pad_to_odd
+
 
 def load_label_to_tensor(label_path):
     label_tensor = tf.py_function(__load_label_to_tensor_internel, [label_path], tf.int32)
@@ -33,3 +35,25 @@ def load_image_tensor_from_path(image_path, label_path=None):
         label_tensor = load_label_to_tensor(label_path)
 
     return image_tensor, label_tensor
+
+
+def simple_load_image(image_path):
+
+    image_tensor, _ = load_image_tensor_from_path(image_path)
+    image_tensor = tf.expand_dims(tf.cast(image_tensor, tf.float32), axis=0)  # [1, H, W, 3]
+
+    image_size = tf.shape(image_tensor)[1:3]
+
+    pad_height = tf.cast(tf.math.ceil(image_size[0] / 32) * 32, tf.int32)
+    pad_width = tf.cast(tf.math.ceil(image_size[1] / 32) * 32, tf.int32)
+
+    pad_height = pad_height if pad_height % 2 != 0 else pad_height + 1
+    pad_width = pad_height if pad_width % 2 != 0 else pad_width + 1
+
+    from iseg.data_process.utils import pad_to_bounding_box, normalize_value_range
+
+    pad_image_tensor = pad_to_bounding_box(image_tensor, 0, 0, pad_height, pad_width, pad_value=[127.5, 127.5, 127.5])
+    pad_image_tensor = normalize_value_range(pad_image_tensor)
+
+    return pad_image_tensor, image_size
+