change transforms/image to work with yx order

chainercv/transforms/image/center_crop.py

@@ -26,9 +26,9 @@ def center_crop(img, size, return_param=False, copy=False):
         contents are listed below with key, value-type and the description
         of the value.
 
-        * **x_slice** (*slice*): A slice used to crop the input image.\
-            The relation below holds together with :obj:`y_slice`.
-        * **y_slice** (*slice*): Similar to :obj:`x_slice`.
+        * **y_slice** (*slice*): A slice used to crop the input image.\
+            The relation below holds together with :obj:`x_slice`.
+        * **x_slice** (*slice*): Similar to :obj:`y_slice`.
 
             .. code::
 
@@ -40,18 +40,18 @@ def center_crop(img, size, return_param=False, copy=False):
     if oW > W or oH > H:
         raise ValueError('shape of image needs to be larger than size')
 
-    x_offset = int(round((W - oW) / 2.))
     y_offset = int(round((H - oH) / 2.))
+    x_offset = int(round((W - oW) / 2.))
 
-    x_slice = slice(x_offset, x_offset + oW)
     y_slice = slice(y_offset, y_offset + oH)
+    x_slice = slice(x_offset, x_offset + oW)
 
     img = img[:, y_slice, x_slice]
 
     if copy:
         img = img.copy()
 
     if return_param:
-        return img, {'x_slice': x_slice, 'y_slice': y_slice}
+        return img, {'y_slice': y_slice, 'x_slice': x_slice}
     else:
         return img

chainercv/transforms/image/flip.py

@@ -1,20 +1,20 @@
-def flip(img, x_flip=False, y_flip=False, copy=False):
+def flip(img, y_flip=False, x_flip=False, copy=False):
     """Flip an image in vertical or horizontal direction as specified.
 
     Args:
         img (~numpy.ndarray): An array that gets flipped. This is in CHW
             format.
-        x_flip (bool): Flip in horizontal direction.
         y_flip (bool): Flip in vertical direction.
+        x_flip (bool): Flip in horizontal direction.
         copy (bool): If False, a view of :obj:`img` will be returned.
 
     Returns:
         Transformed :obj:`img` in CHW format.
     """
-    if x_flip:
-        img = img[:, :, ::-1]
     if y_flip:
         img = img[:, ::-1, :]
+    if x_flip:
+        img = img[:, :, ::-1]
 
     if copy:
         img = img.copy()

chainercv/transforms/image/random_crop.py

@@ -30,9 +30,9 @@ def random_crop(img, size, return_param=False, copy=False):
         contents are listed below with key, value-type and the description
         of the value.
 
-        * **x_slice** (*slice*): A slice used to crop the input image.\
-            The relation below holds together with :obj:`y_slice`.
-        * **y_slice** (*slice*): Similar to :obj:`x_slice`.
+        * **y_slice** (*slice*): A slice used to crop the input image.\
+            The relation below holds together with :obj:`x_slice`.
+        * **x_slice** (*slice*): Similar to :obj:`x_slice`.
 
             .. code::
 
@@ -41,14 +41,6 @@ def random_crop(img, size, return_param=False, copy=False):
     """
     H, W = size
 
-    if img.shape[2] == W:
-        x_offset = 0
-    elif img.shape[2] > W:
-        x_offset = random.choice(six.moves.range(img.shape[2] - W))
-    else:
-        raise ValueError('shape of image needs to be larger than output shape')
-    x_slice = slice(x_offset, x_offset + W)
-
     if img.shape[1] == H:
         y_offset = 0
     elif img.shape[1] > H:
@@ -57,12 +49,20 @@ def random_crop(img, size, return_param=False, copy=False):
         raise ValueError('shape of image needs to be larger than output shape')
     y_slice = slice(y_offset, y_offset + H)
 
+    if img.shape[2] == W:
+        x_offset = 0
+    elif img.shape[2] > W:
+        x_offset = random.choice(six.moves.range(img.shape[2] - W))
+    else:
+        raise ValueError('shape of image needs to be larger than output shape')
+    x_slice = slice(x_offset, x_offset + W)
+
     img = img[:, y_slice, x_slice]
 
     if copy:
         img = img.copy()
 
     if return_param:
-        return img, {'x_slice': x_slice, 'y_slice': y_slice}
+        return img, {'y_slice': y_slice, 'x_slice': x_slice}
     else:
         return img

chainercv/transforms/image/random_expand.py

@@ -39,16 +39,16 @@ def random_expand(img, max_ratio=4, fill=0, return_param=False):
         of the value.
 
         * **ratio** (*float*): The sampled value used to make the canvas.
-        * **x_offset** (*int*): The x coordinate of the top left corner\
-            of the image after placing on the canvas.
         * **y_offset** (*int*): The y coodinate of the top left corner of\
             the image after placing on the canvas.
+        * **x_offset** (*int*): The x coordinate of the top left corner\
+            of the image after placing on the canvas.
 
     """
 
     if max_ratio <= 1:
         if return_param:
-            return img, {'ratio': 1, 'x_offset': 0, 'y_offset': 0}
+            return img, {'ratio': 1, 'y_offset': 0, 'x_offset': 0}
         else:
             return img
 
@@ -57,15 +57,15 @@ def random_expand(img, max_ratio=4, fill=0, return_param=False):
     ratio = random.uniform(1, max_ratio)
     out_H, out_W = int(H * ratio), int(W * ratio)
 
-    x_offset = random.randint(0, out_W - W)
     y_offset = random.randint(0, out_H - H)
+    x_offset = random.randint(0, out_W - W)
 
     out_img = np.empty((C, out_H, out_W), dtype=img.dtype)
     out_img[:] = np.array(fill).reshape(-1, 1, 1)
     out_img[:, y_offset:y_offset + H, x_offset:x_offset + W] = img
 
     if return_param:
-        param = {'ratio': ratio, 'x_offset': x_offset, 'y_offset': y_offset}
+        param = {'ratio': ratio, 'y_offset': y_offset, 'x_offset': x_offset}
         return out_img, param
     else:
         return out_img

chainercv/transforms/image/random_flip.py

@@ -1,15 +1,15 @@
 import random
 
 
-def random_flip(img, x_random=False, y_random=False,
+def random_flip(img, y_random=False, x_random=False,
                 return_param=False, copy=False):
     """Randomly flip an image in vertical or horizontal direction.
 
     Args:
         img (~numpy.ndarray): An array that gets flipped. This is in
             CHW format.
-        x_random (bool): Randomly flip in horizontal direction.
         y_random (bool): Randomly flip in vertical direction.
+        x_random (bool): Randomly flip in horizontal direction.
         return_param (bool): Returns information of flip.
         copy (bool): If False, a view of :obj:`img` will be returned.
 
@@ -25,27 +25,27 @@ def random_flip(img, x_random=False, y_random=False,
         contents are listed below with key, value-type and the description
         of the value.
 
-        * **x_flip** (*bool*): Whether the image was flipped in the\
-            horizontal direction or not.
         * **y_flip** (*bool*): Whether the image was flipped in the\
             vertical direction or not.
+        * **x_flip** (*bool*): Whether the image was flipped in the\
+            horizontal direction or not.
 
     """
     x_flip, y_flip = False, False
-    if x_random:
-        x_flip = random.choice([True, False])
     if y_random:
         y_flip = random.choice([True, False])
+    if x_random:
+        x_flip = random.choice([True, False])
 
-    if x_flip:
-        img = img[:, :, ::-1]
     if y_flip:
         img = img[:, ::-1, :]
+    if x_flip:
+        img = img[:, :, ::-1]
 
     if copy:
         img = img.copy()
 
     if return_param:
-        return img, {'x_flip': x_flip, 'y_flip': y_flip}
+        return img, {'y_flip': y_flip, 'x_flip': x_flip}
     else:
         return img

chainercv/transforms/image/resize_contain.py

@@ -33,10 +33,10 @@ def resize_contain(img, size, fill=0, return_param=False):
         contents are listed below with key, value-type and the description
         of the value.
 
-        * **x_offset** (*int*): The x coordinate of the top left corner\
-            of the image after placing on the canvas.
         * **y_offset** (*int*): The y coodinate of the top left corner of\
             the image after placing on the canvas.
+        * **x_offset** (*int*): The x coordinate of the top left corner\
+            of the image after placing on the canvas.
         * **scaled_size** (*tuple*): The size to which the image is scaled\
             to before placing it on a canvas. This is a tuple of two elements:\
             :obj:`height, width`.
@@ -50,13 +50,13 @@ def resize_contain(img, size, fill=0, return_param=False):
     scaled_size = (int(H * scale), int(W * scale))
     if scale < 1.:
         img = resize(img, scaled_size)
-    x_slice, y_slice = _get_pad_slice(img, size=size)
+    y_slice, x_slice = _get_pad_slice(img, size=size)
     out_img = np.empty((C, out_H, out_W), dtype=img.dtype)
     out_img[:] = np.array(fill).reshape(-1, 1, 1)
     out_img[:, y_slice, x_slice] = img
 
     if return_param:
-        param = {'x_offset': x_slice.start, 'y_offset': y_slice.start,
+        param = {'y_offset': y_slice.start, 'x_offset': x_slice.start,
                  'scaled_size': scaled_size}
         return out_img, param
     else:
@@ -92,4 +92,4 @@ def _get_pad_slice(img, size):
     else:
         x_slice = slice(0, int(size[1]))
 
-    return x_slice, y_slice
+    return y_slice, x_slice

tests/transforms_tests/image_tests/test_center_crop.py

@@ -12,8 +12,8 @@ def test_center_crop(self):
         img = np.random.uniform(size=(3, 48, 32))
 
         out, param = center_crop(img, (24, 16), return_param=True)
-        x_slice = param['x_slice']
         y_slice = param['y_slice']
+        x_slice = param['x_slice']
 
         np.testing.assert_equal(out, img[:, y_slice, x_slice])
         self.assertEqual(x_slice, slice(8, 24))

tests/transforms_tests/image_tests/test_flip_transform.py

@@ -11,7 +11,7 @@ class TestRandomFlip(unittest.TestCase):
     def test_random_flip(self):
         img = np.random.uniform(size=(3, 24, 24))
 
-        out = flip(img, x_flip=True, y_flip=True)
+        out = flip(img, y_flip=True, x_flip=True)
 
         expected = img
         expected = expected[:, :, ::-1]
@@ -21,7 +21,7 @@ def test_random_flip(self):
     def test_random_flip_vertical(self):
         img = np.random.uniform(size=(3, 24, 24))
 
-        out = flip(img, x_flip=False, y_flip=True)
+        out = flip(img, y_flip=True, x_flip=False)
 
         expected = img
         expected = expected[:, ::-1, :]

tests/transforms_tests/image_tests/test_random_crop.py

@@ -12,11 +12,11 @@ def test_random_crop(self):
         img = np.random.uniform(size=(3, 48, 32))
 
         out, param = random_crop(img, (48, 32), return_param=True)
-        x_slice = param['x_slice']
         y_slice = param['y_slice']
+        x_slice = param['x_slice']
         np.testing.assert_equal(out, img)
-        self.assertEqual(x_slice, slice(0, 32))
         self.assertEqual(y_slice, slice(0, 48))
+        self.assertEqual(x_slice, slice(0, 32))
 
         out = random_crop(img, (24, 12))
         self.assertEqual(out.shape[1:], (24, 12))

tests/transforms_tests/image_tests/test_random_expand.py

@@ -23,8 +23,8 @@ def test_random_expand(self):
         out, param = random_expand(
             img, max_ratio=self.max_ratio, return_param=True)
         ratio = param['ratio']
-        x_offset = param['x_offset']
         y_offset = param['y_offset']
+        x_offset = param['x_offset']
         np.testing.assert_equal(
             out[:, y_offset:y_offset + 64, x_offset:x_offset + 32], img)
         self.assertGreaterEqual(ratio, 1)
@@ -47,9 +47,9 @@ def test_random_expand_fill(self):
 
         while True:
             out, param = random_expand(img, fill=self.fill, return_param=True)
-            x_offset = param['x_offset']
             y_offset = param['y_offset']
-            if x_offset > 0 or y_offset > 0:
+            x_offset = param['x_offset']
+            if y_offset > 0 or x_offset > 0:
                 break
 
         if isinstance(self.fill, int):

tests/transforms_tests/image_tests/test_random_flip_transform.py

@@ -12,15 +12,15 @@ def test_random_flip(self):
         img = np.random.uniform(size=(3, 24, 24))
 
         out, param = random_flip(
-            img, x_random=True, y_random=True, return_param=True)
-        x_flip = param['x_flip']
+            img, y_random=True, x_random=True, return_param=True)
         y_flip = param['y_flip']
+        x_flip = param['x_flip']
 
         expected = img
-        if x_flip:
-            expected = expected[:, :, ::-1]
         if y_flip:
             expected = expected[:, ::-1, :]
+        if x_flip:
+            expected = expected[:, :, ::-1]
         np.testing.assert_equal(out, expected)
 
 

tests/transforms_tests/image_tests/test_resize_contain.py

@@ -22,26 +22,26 @@ def test_resize_contain(self):
         np.testing.assert_array_equal(img, out[:, 8:40, 16:80])
         np.testing.assert_array_equal(self.fill, out[:, 0, 0])
         self.assertEqual(param['scaled_size'], (32, 64))
-        self.assertEqual(param['x_offset'], 16)
         self.assertEqual(param['y_offset'], 8)
+        self.assertEqual(param['x_offset'], 16)
 
     def test_resize_contain_canvas_small_x(self):
         img = np.random.uniform(size=(3, 32, 64))
 
         out, param = resize_contain(
             img, (16, 68), fill=self.fill, return_param=True)
         self.assertEqual(param['scaled_size'], (16, 32))
-        self.assertEqual(param['x_offset'], 18)
         self.assertEqual(param['y_offset'], 0)
+        self.assertEqual(param['x_offset'], 18)
 
     def test_resize_contain_canvas_small_y(self):
         img = np.random.uniform(size=(3, 32, 64))
 
         out, param = resize_contain(
             img, (24, 16), fill=self.fill, return_param=True)
         self.assertEqual(param['scaled_size'], (8, 16))
-        self.assertEqual(param['x_offset'], 0)
         self.assertEqual(param['y_offset'], 8)
+        self.assertEqual(param['x_offset'], 0)
 
 
 testing.run_module(__name__, __file__)