ColorNormalizeAug now accepts mxnet NDArrays as well as stated (wrong…

…ly) in the documentation. (apache#11606)

* ColorNormalizeAug now accepts mxnet NDArrays as well.

According to the doc of mx.image.ColorNormalizeAug mean and std must be mxnet NDArrays though in reality it accepts only numpy arrays. This commit allows ColorNormalizeAug to accept both numpy and mxnet ndarrays.

Extended the image unit tests to cover this functionality.

* Fixed missing brackets

* Simplified assert call according to pylint

* Fixed linting issue

* Fixed namespace issue

* Fixed std and mean assignment

* Fixed isinstance python3 mode

* Added with_seed annotation

CONTRIBUTORS.md

@@ -173,3 +173,4 @@ List of Contributors
 * [Jesse Brizzi](https://github.com/jessebrizzi)
 * [Hang Zhang](http://hangzh.com)
 * [Kou Ding](https://github.com/chinakook)
+* [Istvan Fehervari](https://github.com/ifeherva)

python/mxnet/image/image.py

@@ -839,8 +839,8 @@ class ColorNormalizeAug(Augmenter):
     """
     def __init__(self, mean, std):
         super(ColorNormalizeAug, self).__init__(mean=mean, std=std)
-        self.mean = nd.array(mean) if mean is not None else None
-        self.std = nd.array(std) if std is not None else None
+        self.mean = mean if mean is None or isinstance(mean, nd.NDArray) else nd.array(mean)
+        self.std = std if std is None or isinstance(std, nd.NDArray) else nd.array(std)
 
     def __call__(self, src):
         """Augmenter body"""
@@ -999,14 +999,14 @@ def CreateAugmenter(data_shape, resize=0, rand_crop=False, rand_resize=False, ra
         auglist.append(RandomGrayAug(rand_gray))
 
     if mean is True:
-        mean = np.array([123.68, 116.28, 103.53])
+        mean = nd.array([123.68, 116.28, 103.53])
     elif mean is not None:
-        assert isinstance(mean, np.ndarray) and mean.shape[0] in [1, 3]
+        assert isinstance(mean, (np.ndarray, nd.NDArray)) and mean.shape[0] in [1, 3]
 
     if std is True:
-        std = np.array([58.395, 57.12, 57.375])
+        std = nd.array([58.395, 57.12, 57.375])
     elif std is not None:
-        assert isinstance(std, np.ndarray) and std.shape[0] in [1, 3]
+        assert isinstance(std, (np.ndarray, nd.NDArray)) and std.shape[0] in [1, 3]
 
     if mean is not None or std is not None:
         auglist.append(ColorNormalizeAug(mean, std))

tests/python/unittest/test_image.py

@@ -18,7 +18,7 @@
 import mxnet as mx
 import numpy as np
 from mxnet.test_utils import *
-from common import assertRaises
+from common import assertRaises, with_seed
 import shutil
 import tempfile
 import unittest
@@ -153,8 +153,19 @@ def test_imageiter(self):
         for batch in test_iter:
             pass
 
-
+    @with_seed()
     def test_augmenters(self):
+        # ColorNormalizeAug
+        mean = np.random.rand(3) * 255
+        std = np.random.rand(3) + 1
+        width = np.random.randint(100, 500)
+        height = np.random.randint(100, 500)
+        src = np.random.rand(height, width, 3) * 255.
+        # We test numpy and mxnet NDArray inputs
+        color_norm_aug = mx.image.ColorNormalizeAug(mean=mx.nd.array(mean), std=std)
+        out_image = color_norm_aug(mx.nd.array(src))
+        assert_almost_equal(out_image.asnumpy(), (src - mean) / std, atol=1e-3)
+
         # only test if all augmenters will work
         # TODO(Joshua Zhang): verify the augmenter outputs
         im_list = [[0, x] for x in TestImage.IMAGES]