refactor: Avoid creating new transformation objects with each call

transformations/object_detection_2d_photometric_ops.py

@@ -78,7 +78,7 @@ def __init__(self, to='uint8'):
 
  def __call__(self, image, labels=None):
  if self.to == 'uint8':
- image = image.astype(np.uint8)
+ image = np.round(image, decimals=0).astype(np.uint8)
  else:
  image = image.astype(np.float32)
  if labels is None:
@@ -152,13 +152,13 @@ def __init__(self, max_delta=18, prob=0.5):
  if not (0 <= max_delta <= 180): raise ValueError("`max_delta` must be in the closed interval `[0, 180]`.")
  self.max_delta = max_delta
  self.prob = prob
+ self.change_hue = Hue(delta=0)
 
  def __call__(self, image, labels=None):
  p = np.random.uniform(0,1)
  if p >= (1.0-self.prob):
- delta = np.random.uniform(-self.max_delta, self.max_delta)
- change_hue = Hue(delta=delta)
- return change_hue(image, labels)
+ self.change_hue.delta = np.random.uniform(-self.max_delta, self.max_delta)
+ return self.change_hue(image, labels)
  elif labels is None:
  return image
  else:
@@ -211,13 +211,13 @@ def __init__(self, lower=0.3, upper=2.0, prob=0.5):
  self.lower = lower
  self.upper = upper
  self.prob = prob
+ self.change_saturation = Saturation(factor=1.0)
 
  def __call__(self, image, labels=None):
  p = np.random.uniform(0,1)
  if p >= (1.0-self.prob):
- factor = np.random.uniform(self.lower, self.upper)
- change_saturation = Saturation(factor=factor)
- return change_saturation(image, labels)
+ self.change_saturation.factor = np.random.uniform(self.lower, self.upper)
+ return self.change_saturation(image, labels)
  elif labels is None:
  return image
  else:
@@ -267,13 +267,13 @@ def __init__(self, lower=-84, upper=84, prob=0.5):
  self.lower = float(lower)
  self.upper = float(upper)
  self.prob = prob
+ self.change_brightness = Brightness(delta=0)
 
  def __call__(self, image, labels=None):
  p = np.random.uniform(0,1)
  if p >= (1.0-self.prob):
- delta = np.random.uniform(self.lower, self.upper)
- change_brightness = Brightness(delta=delta)
- return change_brightness(image, labels)
+ self.change_brightness.delta = np.random.uniform(self.lower, self.upper)
+ return self.change_brightness(image, labels)
  elif labels is None:
  return image
  else:
@@ -326,13 +326,13 @@ def __init__(self, lower=0.5, upper=1.5, prob=0.5):
  self.lower = lower
  self.upper = upper
  self.prob = prob
+ self.change_contrast = Contrast(factor=1.0)
 
  def __call__(self, image, labels=None):
  p = np.random.uniform(0,1)
  if p >= (1.0-self.prob):
- factor = np.random.uniform(self.lower, self.upper)
- change_contrast = Contrast(factor=factor)
- return change_contrast(image, labels)
+ self.change_contrast.factor = np.random.uniform(self.lower, self.upper)
+ return self.change_contrast(image, labels)
  elif labels is None:
  return image
  else:
@@ -425,12 +425,12 @@ def __init__(self, prob=0.5):
  unaltered image is returned.
  '''
  self.prob = prob
+ self.equalize = HistogramEqualization()
 
  def __call__(self, image, labels=None):
  p = np.random.uniform(0,1)
  if p >= (1.0-self.prob):
- equalize = HistogramEqualization()
- return equalize(image, labels)
+ return self.equalize(image, labels)
  elif labels is None:
  return image
  else:
@@ -472,14 +472,14 @@ def __init__(self, prob=0.5):
  self.permutations = ((0, 2, 1),
  (1, 0, 2), (1, 2, 0),
  (2, 0, 1), (2, 1, 0))
+ self.swap_channels = ChannelSwap(order=(0, 1, 2))
 
  def __call__(self, image, labels=None):
  p = np.random.uniform(0,1)
  if p >= (1.0-self.prob):
  i = np.random.randint(5) # There are 6 possible permutations.
- permutation = self.permutations[i]
- swap_channels = ChannelSwap(order=permutation)
- return swap_channels(image, labels)
+ self.swap_channels.order = self.permutations[i]
+ return self.swap_channels(image, labels)
  elif labels is None:
  return image
  else: