Python tensorflow.python.framework.dtypes.uint8() Examples

def extract_labels(dataframe, one_hot=False, num_classes=10):
    """Extract the labels into a 1D uint8 numpy array [index].

    Args:
    dataframe: A pandas dataframe object.
    one_hot: Does one hot encoding for the result.
    num_classes: Number of classes for the one hot encoding.

    Returns:
    labels: a 1D uint8 numpy array.
    """
    print('Extracting labels', )
    labels = dataframe['label'].values
    labels = _label_encoder.fit_transform(labels)
    if one_hot:
        return dense_to_one_hot(labels, num_classes)
    return labels 

def testResizeDownArea(self):
    img_shape = [1, 6, 6, 1]
    data = [128, 64, 32, 16, 8, 4,
            4, 8, 16, 32, 64, 128,
            128, 64, 32, 16, 8, 4,
            5, 10, 15, 20, 25, 30,
            30, 25, 20, 15, 10, 5,
            5, 10, 15, 20, 25, 30]
    img_np = np.array(data, dtype=np.uint8).reshape(img_shape)

    target_height = 4
    target_width = 4
    expected_data = [73, 33, 23, 39,
                     73, 33, 23, 39,
                     14, 16, 19, 21,
                     14, 16, 19, 21]

    with self.test_session(use_gpu=True):
      image = constant_op.constant(img_np, shape=img_shape)
      y = image_ops.resize_images(image, [target_height, target_width],
                                  image_ops.ResizeMethod.AREA)
      expected = np.array(expected_data).reshape(
          [1, target_height, target_width, 1])
      resized = y.eval()
      self.assertAllClose(resized, expected, atol=1) 

def GenerateImage(self, image_format, image_shape):
    """Generates an image and an example containing the encoded image.

    Args:
      image_format: the encoding format of the image.
      image_shape: the shape of the image to generate.

    Returns:
      image: the generated image.
      example: a TF-example with a feature key 'image/encoded' set to the
        serialized image and a feature key 'image/format' set to the image
        encoding format ['jpeg', 'JPEG', 'png', 'PNG', 'raw'].
    """
    num_pixels = image_shape[0] * image_shape[1] * image_shape[2]
    image = np.linspace(
        0, num_pixels - 1, num=num_pixels).reshape(image_shape).astype(np.uint8)
    tf_encoded = self._Encoder(image, image_format)
    example = example_pb2.Example(features=feature_pb2.Features(feature={
        'image/encoded': self._EncodedBytesFeature(tf_encoded),
        'image/format': self._StringFeature(image_format)
    }))

    return image, example.SerializeToString() 

def _RGBToGrayscale(self, images):
    is_batch = True
    if len(images.shape) == 3:
      is_batch = False
      images = np.expand_dims(images, axis=0)
    out_shape = images.shape[0:3] + (1,)
    out = np.zeros(shape=out_shape, dtype=np.uint8)
    for batch in xrange(images.shape[0]):
      for y in xrange(images.shape[1]):
        for x in xrange(images.shape[2]):
          red = images[batch, y, x, 0]
          green = images[batch, y, x, 1]
          blue = images[batch, y, x, 2]
          gray = 0.2989 * red + 0.5870 * green + 0.1140 * blue
          out[batch, y, x, 0] = int(gray)
    if not is_batch:
      out = np.squeeze(out, axis=0)
    return out 

def testConvertBetweenInt16AndInt8(self):
    with self.test_session(use_gpu=True):
      # uint8, uint16
      self._convert([0, 255 * 256], dtypes.uint16, dtypes.uint8,
                    [0, 255])
      self._convert([0, 255], dtypes.uint8, dtypes.uint16,
                    [0, 255 * 256])
      # int8, uint16
      self._convert([0, 127 * 2 * 256], dtypes.uint16, dtypes.int8,
                    [0, 127])
      self._convert([0, 127], dtypes.int8, dtypes.uint16,
                    [0, 127 * 2 * 256])
      # int16, uint16
      self._convert([0, 255 * 256], dtypes.uint16, dtypes.int16,
                    [0, 255 * 128])
      self._convert([0, 255 * 128], dtypes.int16, dtypes.uint16,
                    [0, 255 * 256]) 

def testBasicGrayscaleToRGB(self):
    # 4-D input with batch dimension.
    x_np = np.array([[1, 2]], dtype=np.uint8).reshape([1, 1, 2, 1])
    y_np = np.array([[1, 1, 1], [2, 2, 2]],
                    dtype=np.uint8).reshape([1, 1, 2, 3])

    with self.test_session(use_gpu=True):
      x_tf = constant_op.constant(x_np, shape=x_np.shape)
      y = image_ops.grayscale_to_rgb(x_tf)
      y_tf = y.eval()
      self.assertAllEqual(y_tf, y_np)

    # 3-D input with no batch dimension.
    x_np = np.array([[1, 2]], dtype=np.uint8).reshape([1, 2, 1])
    y_np = np.array([[1, 1, 1], [2, 2, 2]], dtype=np.uint8).reshape([1, 2, 3])

    with self.test_session(use_gpu=True):
      x_tf = constant_op.constant(x_np, shape=x_np.shape)
      y = image_ops.grayscale_to_rgb(x_tf)
      y_tf = y.eval()
      self.assertAllEqual(y_tf, y_np) 

def visualize(self):
    """Multi-channel visualization of densities as images.

    Creates and returns an image summary visualizing the current probabilty
    density estimates. The image contains one row for each channel. Within each
    row, the pixel intensities are proportional to probability values, and each
    row is centered on the median of the corresponding distribution.

    Returns:
      The created image summary.
    """
    with ops.name_scope(self._name_scope()):
      image = self._pmf
      image *= 255 / math_ops.reduce_max(image, axis=1, keepdims=True)
      image = math_ops.cast(image + .5, dtypes.uint8)
      image = image[None, :, :, None]
    return summary.image("pmf", image, max_outputs=1) 

def test_adjust_gamma_less_one(self):
    """Verifying the output with expected results for gamma
    correction with gamma equal to half"""
    with self.test_session():
      x_np = np.arange(0, 255, 4, np.uint8).reshape(8,8)
      y = image_ops.adjust_gamma(x_np, gamma=0.5)
      y_tf = np.trunc(y.eval())

      y_np = np.array([[  0,  31,  45,  55,  63,  71,  78,  84],
          [ 90,  95, 100, 105, 110, 115, 119, 123],
          [127, 131, 135, 139, 142, 146, 149, 153],
          [156, 159, 162, 165, 168, 171, 174, 177],
          [180, 183, 186, 188, 191, 194, 196, 199],
          [201, 204, 206, 209, 211, 214, 216, 218],
          [221, 223, 225, 228, 230, 232, 234, 236],
          [238, 241, 243, 245, 247, 249, 251, 253]], dtype=np.float32)

      self.assertAllClose(y_tf, y_np, 1e-6) 

def test_adjust_gamma_greater_one(self):
    """Verifying the output with expected results for gamma
    correction with gamma equal to two"""
    with self.test_session():
      x_np = np.arange(0, 255, 4, np.uint8).reshape(8,8)
      y = image_ops.adjust_gamma(x_np, gamma=2)
      y_tf = np.trunc(y.eval())

      y_np = np.array([[  0,   0,   0,   0,   1,   1,   2,   3],
          [  4,   5,   6,   7,   9,  10,  12,  14],
          [ 16,  18,  20,  22,  25,  27,  30,  33],
          [ 36,  39,  42,  45,  49,  52,  56,  60],
          [ 64,  68,  72,  76,  81,  85,  90,  95],
          [100, 105, 110, 116, 121, 127, 132, 138],
          [144, 150, 156, 163, 169, 176, 182, 189],
          [196, 203, 211, 218, 225, 233, 241, 249]], dtype=np.float32)

      self.assertAllClose(y_tf, y_np, 1e-6) 

def extract_labels(f, one_hot=False, num_classes=10):
  """Extract the labels into a 1D uint8 numpy array [index].

  Args:
    f: A file object that can be passed into a gzip reader.
    one_hot: Does one hot encoding for the result.
    num_classes: Number of classes for the one hot encoding.

  Returns:
    labels: a 1D uint8 numpy array.

  Raises:
    ValueError: If the bystream doesn't start with 2049.
  """
  print('Extracting', f.name)
  with gzip.GzipFile(fileobj=f) as bytestream:
    magic = _read32(bytestream)
    if magic != 2049:
      raise ValueError('Invalid magic number %d in MNIST label file: %s' %
                       (magic, f.name))
    num_items = _read32(bytestream)
    buf = bytestream.read(num_items)
    labels = numpy.frombuffer(buf, dtype=numpy.uint8)
    if one_hot:
      return dense_to_one_hot(labels, num_classes)
    return labels 

def extract_images(f):
  """Extract the images into a 4D uint8 numpy array [index, y, x, depth].

  Args:
    f: A file object that can be passed into a gzip reader.

  Returns:
    data: A 4D uint8 numpy array [index, y, x, depth].

  Raises:
    ValueError: If the bytestream does not start with 2051.

  """
  print('Extracting', f.name)
  with gzip.GzipFile(fileobj=f) as bytestream:
    magic = _read32(bytestream)
    if magic != 2051:
      raise ValueError('Invalid magic number %d in MNIST image file: %s' %
                       (magic, f.name))
    num_images = _read32(bytestream)
    rows = _read32(bytestream)
    cols = _read32(bytestream)
    buf = bytestream.read(rows * cols * num_images)
    data = numpy.frombuffer(buf, dtype=numpy.uint8)
    data = data.reshape(num_images, rows, cols, 1)
    return data 

def extract_labels(f, one_hot=False, num_classes=10):
  """Extract the labels into a 1D uint8 numpy array [index].

  Args:
    f: A file object that can be passed into a gzip reader.
    one_hot: Does one hot encoding for the result.
    num_classes: Number of classes for the one hot encoding.

  Returns:
    labels: a 1D uint8 numpy array.

  Raises:
    ValueError: If the bystream doesn't start with 2049.
  """
  print('Extracting', f.name)
  with gzip.GzipFile(fileobj=f) as bytestream:
    magic = _read32(bytestream)
    if magic != 2049:
      raise ValueError('Invalid magic number %d in MNIST label file: %s' %
                       (magic, f.name))
    num_items = _read32(bytestream)
    buf = bytestream.read(num_items)
    labels = numpy.frombuffer(buf, dtype=numpy.uint8)
    if one_hot:
      return dense_to_one_hot(labels, num_classes)
    return labels 

def __init__(self,
               images,
               labels,
               fake_data=False,
               one_hot=False,
               dtype=dtypes.float32,
               reshape=True):
    """Construct a DataSet.
    one_hot arg is used only if fake_data is true.  `dtype` can be either
    `uint8` to leave the input as `[0, 255]`, or `float32` to rescale into
    `[0, 1]`.
    """
    dtype = dtypes.as_dtype(dtype).base_dtype
    if dtype not in (dtypes.uint8, dtypes.float32):
      raise TypeError('Invalid image dtype %r, expected uint8 or float32' %
                      dtype)
    if fake_data:
      self._num_examples = 10000
      self.one_hot = one_hot
    else:
      assert images.shape[0] == labels.shape[0], (
          'images.shape: %s labels.shape: %s' % (images.shape, labels.shape))
      self._num_examples = images.shape[0]

      # Convert shape from [num examples, rows, columns, depth]
      # to [num examples, rows*columns] (assuming depth == 1)
      if reshape:
        assert images.shape[3] == 1
        images = images.reshape(images.shape[0],
                                images.shape[1] * images.shape[2])
      if dtype == dtypes.float32:
        # Convert from [0, 255] -> [0.0, 1.0].
        images = images.astype(numpy.float32)
        images = numpy.multiply(images, 1.0 / 255.0)
    self._images = images
    self._labels = labels
    self._epochs_completed = 0
    self._index_in_epoch = 0 

def extract_images(f):
  """Extract the images into a 4D uint8 numpy array [index, y, x, depth].

  Args:
    f: A file object that can be passed into a gzip reader.

  Returns:
    data: A 4D uint8 numpy array [index, y, x, depth].

  Raises:
    ValueError: If the bytestream does not start with 2051.

  """
  print('Extracting', f.name)
  with gzip.GzipFile(fileobj=f) as bytestream:
    magic = _read32(bytestream)
    if magic != 2051:
      raise ValueError('Invalid magic number %d in MNIST image file: %s' %
                       (magic, f.name))
    num_images = _read32(bytestream)
    rows = _read32(bytestream)
    cols = _read32(bytestream)
    buf = bytestream.read(rows * cols * num_images)
    data = numpy.frombuffer(buf, dtype=numpy.uint8)
    data = data.reshape(num_images, rows, cols, 1)
    return data 

def __init__(self,
               images,
               labels,
               fake_data=False,
               one_hot=False,
               dtype=dtypes.float32,
               reshape=True):
    """Construct a DataSet.
    one_hot arg is used only if fake_data is true.  `dtype` can be either
    `uint8` to leave the input as `[0, 255]`, or `float32` to rescale into
    `[0, 1]`.
    """
    dtype = dtypes.as_dtype(dtype).base_dtype
    if dtype not in (dtypes.uint8, dtypes.float32):
      raise TypeError('Invalid image dtype %r, expected uint8 or float32' %
                      dtype)
    if fake_data:
      self._num_examples = 10000
      self.one_hot = one_hot
    else:
      assert images.shape[0] == labels.shape[0], (
          'images.shape: %s labels.shape: %s' % (images.shape, labels.shape))
      self._num_examples = images.shape[0]

    if dtype == dtypes.float32:
        images = images.astype(numpy.float32)

    self._images = images
    self._labels = labels
    self._epochs_completed = 0
    self._index_in_epoch = 0 

def testNoConvert(self):
    # Make sure converting to the same data type creates only an identity op
    with self.test_session(use_gpu=True):
      image = constant_op.constant([1], dtype=dtypes.uint8)
      image_ops.convert_image_dtype(image, dtypes.uint8)
      y = image_ops.convert_image_dtype(image, dtypes.uint8)
      self.assertEquals(y.op.type, 'Identity')
      self.assertEquals(y.op.inputs[0], image) 

def testPartialShapes(self):
    p_unknown_rank = array_ops.placeholder(dtypes.uint8)
    p_unknown_dims = array_ops.placeholder(dtypes.uint8,
                                           shape=[None, None, None])
    p_unknown_width = array_ops.placeholder(dtypes.uint8, shape=[64, None, 3])

    p_wrong_rank = array_ops.placeholder(dtypes.uint8, shape=[None, None])
    p_zero_dim = array_ops.placeholder(dtypes.uint8, shape=[64, 0, 3])

    for op in [image_ops.flip_left_right,
               image_ops.flip_up_down,
               image_ops.random_flip_left_right,
               image_ops.random_flip_up_down,
               image_ops.transpose_image,
               image_ops.rot90]:
      transformed_unknown_rank = op(p_unknown_rank)
      self.assertEqual(3, transformed_unknown_rank.get_shape().ndims)
      transformed_unknown_dims = op(p_unknown_dims)
      self.assertEqual(3, transformed_unknown_dims.get_shape().ndims)
      transformed_unknown_width = op(p_unknown_width)
      self.assertEqual(3, transformed_unknown_width.get_shape().ndims)

      with self.assertRaisesRegexp(ValueError, 'must be three-dimensional'):
        op(p_wrong_rank)
      with self.assertRaisesRegexp(ValueError, 'must be > 0'):
        op(p_zero_dim) 

def __init__(self,
                 payloads,
                 labels,
                 dtype=dtypes.float32,
                 seed=None):
        """Construct a DataSet.
        one_hot arg is used only if fake_data is true.  `dtype` can be either
        `uint8` to leave the input as `[0, 255]`, or `float32` to rescale into
        `[0, 1]`.  Seed arg provides for convenient deterministic testing.
        """
        seed1, seed2 = random_seed.get_seed(seed)
        # If op level seed is not set, use whatever graph level seed is returned
        np.random.seed(seed1 if seed is None else seed2)
        dtype = dtypes.as_dtype(dtype).base_dtype
        if dtype not in (dtypes.uint8, dtypes.float32):
            raise TypeError('Invalid payload dtype %r, expected uint8 or float32' %
                            dtype)

        assert payloads.shape[0] == labels.shape[0], (
                'payloads.shape: %s labels.shape: %s' % (payloads.shape, labels.shape))
        self._num_examples = payloads.shape[0]

        if dtype == dtypes.float32:
            # Convert from [0, 255] -> [0.0, 1.0].
            payloads = payloads.astype(np.float32)
            payloads = np.multiply(payloads, 1.0 / 255.0)

        self._payloads = payloads
        self._labels = labels
        self._epochs_completed = 0
        self._index_in_epoch = 0 

def testValid(self):
    # Read some real GIFs
    prefix = 'tensorflow/core/lib/gif/testdata/'
    filename = 'scan.gif'
    WIDTH = 20
    HEIGHT = 40
    STRIDE = 5
    shape = (12, HEIGHT, WIDTH, 3)

    with self.test_session(use_gpu=True) as sess:
      gif0 = io_ops.read_file(prefix + filename)
      image0 = image_ops.decode_gif(gif0)
      gif0, image0 = sess.run([gif0, image0])

      self.assertEqual(image0.shape, shape)

      for frame_idx, frame in enumerate(image0):
        gt = np.zeros(shape[1:], dtype=np.uint8)
        start = frame_idx * STRIDE
        end = (frame_idx + 1) * STRIDE
        print(frame_idx)
        if end <= WIDTH:
          gt[:, start:end, :] = 255
        else:
          start -= WIDTH
          end -= WIDTH
          gt[start:end, :, :] = 255

        self.assertAllClose(frame, gt) 

def _SimpleColorRamp():
  """Build a simple color ramp RGB image."""
  w, h = 256, 200
  i = np.arange(h)[:, None]
  j = np.arange(w)
  image = np.empty((h, w, 3), dtype=np.uint8)
  image[:, :, 0] = i
  image[:, :, 1] = j
  image[:, :, 2] = (i + j) >> 1
  return image 

def testSumTensor(self):
    img_shape = [1, 6, 4, 1]
    # This test is also conducted with int8, so 127 is the maximum
    # value that can be used.
    data = [127, 127, 64, 64,
            127, 127, 64, 64,
            64, 64, 127, 127,
            64, 64, 127, 127,
            50, 50, 100, 100,
            50, 50, 100, 100]
    # Test size where width is specified as a tensor which is a sum
    # of two tensors.
    width_1 = constant_op.constant(1)
    width_2 = constant_op.constant(3)
    width = math_ops.add(width_1, width_2)
    height = constant_op.constant(6)

    img_np = np.array(data, dtype=np.uint8).reshape(img_shape)

    for opt in self.OPTIONS:
      with self.test_session() as sess:
        image = constant_op.constant(img_np, shape=img_shape)
        y = image_ops.resize_images(image, [height, width], opt)
        yshape = array_ops.shape(y)
        resized, newshape = sess.run([y, yshape])
        self.assertAllEqual(img_shape, newshape)
        self.assertAllClose(resized, img_np, atol=1e-5) 

def testNegativeDelta(self):
    x_shape = [2, 2, 3]
    x_data = [0, 5, 13, 54, 135, 226, 37, 8, 234, 90, 255, 1]
    x_np = np.array(x_data, dtype=np.uint8).reshape(x_shape)

    y_data = [0, 0, 3, 44, 125, 216, 27, 0, 224, 80, 245, 0]
    y_np = np.array(y_data, dtype=np.uint8).reshape(x_shape)

    self._testBrightness(x_np, y_np, delta=-10. / 255.) 

def testPositiveDeltaUint8(self):
    x_shape = [2, 2, 3]
    x_data = [0, 5, 13, 54, 135, 226, 37, 8, 234, 90, 255, 1]
    x_np = np.array(x_data, dtype=np.uint8).reshape(x_shape)

    y_data = [10, 15, 23, 64, 145, 236, 47, 18, 244, 100, 255, 11]
    y_np = np.array(y_data, dtype=np.uint8).reshape(x_shape)

    self._testBrightness(x_np, y_np, delta=10. / 255.) 

def testBatchDoubleContrast(self):
    x_shape = [2, 1, 2, 3]
    x_data = [0, 5, 13, 54, 135, 226, 37, 8, 234, 90, 255, 1]
    x_np = np.array(x_data, dtype=np.uint8).reshape(x_shape)

    y_data = [0, 0, 0, 81, 200, 255, 10, 0, 255, 116, 255, 0]
    y_np = np.array(y_data, dtype=np.uint8).reshape(x_shape)

    self._testContrast(x_np, y_np, contrast_factor=2.0) 

def testHalfContrastUint8(self):
    x_shape = [1, 2, 2, 3]
    x_data = [0, 5, 13, 54, 135, 226, 37, 8, 234, 90, 255, 1]
    x_np = np.array(x_data, dtype=np.uint8).reshape(x_shape)

    y_data = [22, 52, 65, 49, 118, 172, 41, 54, 176, 67, 178, 59]
    y_np = np.array(y_data, dtype=np.uint8).reshape(x_shape)

    self._testContrast(x_np, y_np, contrast_factor=0.5) 

def testRandomUpDown(self):
    x_np = np.array([0, 1], dtype=np.uint8).reshape([2, 1, 1])
    num_iterations = 500

    hist = [0, 0]
    with self.test_session(use_gpu=True):
      x_tf = constant_op.constant(x_np, shape=x_np.shape)
      y = image_ops.random_flip_up_down(x_tf)
      for _ in xrange(num_iterations):
        y_np = y.eval().flatten()[0]
        hist[y_np] += 1

    # Ensure that each entry is observed within 4 standard deviations.
    four_stddev = 4.0 * np.sqrt(num_iterations / 2.0)
    self.assertAllClose(hist, [num_iterations / 2.0] * 2, atol=four_stddev) 

def testRandomLeftRight(self):
    x_np = np.array([0, 1], dtype=np.uint8).reshape([1, 2, 1])
    num_iterations = 500

    hist = [0, 0]
    with self.test_session(use_gpu=True):
      x_tf = constant_op.constant(x_np, shape=x_np.shape)
      y = image_ops.random_flip_left_right(x_tf)
      for _ in xrange(num_iterations):
        y_np = y.eval().flatten()[0]
        hist[y_np] += 1

    # Ensure that each entry is observed within 4 standard deviations.
    four_stddev = 4.0 * np.sqrt(num_iterations / 2.0)
    self.assertAllClose(hist, [num_iterations / 2.0] * 2, atol=four_stddev) 

def testRot90NumpyEquivalence(self):
    image = np.arange(24, dtype=np.uint8).reshape([2, 4, 3])
    with self.test_session(use_gpu=True):
      k_placeholder = array_ops.placeholder(dtypes.int32, shape=[])
      y_tf = image_ops.rot90(image, k_placeholder)
      for k in xrange(4):
        y_np = np.rot90(image, k=k)
        self.assertAllEqual(y_np, y_tf.eval({k_placeholder: k})) 

def testRot90GroupOrder(self):
    image = np.arange(24, dtype=np.uint8).reshape([2, 4, 3])
    with self.test_session(use_gpu=True):
      rotated = image
      for _ in xrange(4):
        rotated = image_ops.rot90(rotated)
      self.assertAllEqual(image, rotated.eval()) 

def __init__(self,
               images,
               labels,
               fake_data=False,
               one_hot=False,
               dtype=dtypes.float32,
               reshape=True):
    """Construct a DataSet.
    one_hot arg is used only if fake_data is true.  `dtype` can be either
    `uint8` to leave the input as `[0, 255]`, or `float32` to rescale into
    `[0, 1]`.
    """
    dtype = dtypes.as_dtype(dtype).base_dtype
    if dtype not in (dtypes.uint8, dtypes.float32):
      raise TypeError('Invalid image dtype %r, expected uint8 or float32' %
                      dtype)
    if fake_data:
      self._num_examples = 10000
      self.one_hot = one_hot
    else:
      assert images.shape[0] == labels.shape[0], (
          'images.shape: %s labels.shape: %s' % (images.shape, labels.shape))
      self._num_examples = images.shape[0]

      # Convert shape from [num examples, rows, columns, depth]
      # to [num examples, rows*columns] (assuming depth == 1)
      if reshape:
        assert images.shape[3] == 1
        images = images.reshape(images.shape[0],
                                images.shape[1] * images.shape[2])
      if dtype == dtypes.float32:
        # Convert from [0, 255] -> [0.0, 1.0].
        images = images.astype(numpy.float32)
        images = numpy.multiply(images, 1.0 / 255.0)
    self._images = images
    self._labels = labels
    self._epochs_completed = 0
    self._index_in_epoch = 0