Python object_detection.core.box_list_ops.scale() Examples

def __new__(cls,
              localization_loss,
              scale_loss_weight,
              offset_loss_weight,
              task_loss_weight=1.0):
    """Constructor with default values for ObjectDetectionParams.

    Args:
      localization_loss: a object_detection.core.losses.Loss object to compute
        the loss for the center offset and height/width predictions in
        CenterNet.
      scale_loss_weight: float, The weight for localizing box size. Note that
        the scale loss is dependent on the input image size, since we penalize
        the raw height and width. This constant may need to be adjusted
        depending on the input size.
      offset_loss_weight: float, The weight for localizing center offsets.
      task_loss_weight: float, the weight of the object detection loss.

    Returns:
      An initialized ObjectDetectionParams namedtuple.
    """
    return super(ObjectDetectionParams,
                 cls).__new__(cls, localization_loss, scale_loss_weight,
                              offset_loss_weight, task_loss_weight) 

def preprocess(self, inputs):
    """Converts a batch of unscaled images to a scale suitable for the model.

    This method normalizes the image using the given `channel_means` and
    `channels_stds` values at initialization time while optionally flipping
    the channel order if `bgr_ordering` is set.

    Args:
      inputs: a [batch, height, width, channels] float32 tensor

    Returns:
      outputs: a [batch, height, width, channels] float32 tensor

    """

    if self._bgr_ordering:
      red, green, blue = tf.unstack(inputs, axis=3)
      inputs = tf.stack([blue, green, red], axis=3)

    channel_means = tf.reshape(tf.constant(self._channel_means),
                               [1, 1, 1, -1])
    channel_stds = tf.reshape(tf.constant(self._channel_stds),
                              [1, 1, 1, -1])

    return (inputs - channel_means)/channel_stds 

def scale_boxes_to_pixel_coordinates(image, boxes, keypoints=None):
  """Scales boxes from normalized to pixel coordinates.

  Args:
    image: A 3D float32 tensor of shape [height, width, channels].
    boxes: A 2D float32 tensor of shape [num_boxes, 4] containing the bounding
      boxes in normalized coordinates. Each row is of the form
      [ymin, xmin, ymax, xmax].
    keypoints: (optional) rank 3 float32 tensor with shape
      [num_instances, num_keypoints, 2]. The keypoints are in y-x normalized
      coordinates.

  Returns:
    image: unchanged input image.
    scaled_boxes: a 2D float32 tensor of shape [num_boxes, 4] containing the
      bounding boxes in pixel coordinates.
    scaled_keypoints: a 3D float32 tensor with shape
      [num_instances, num_keypoints, 2] containing the keypoints in pixel
      coordinates.
  """
  boxlist = box_list.BoxList(boxes)
  image_height = tf.shape(image)[0]
  image_width = tf.shape(image)[1]
  scaled_boxes = box_list_ops.scale(boxlist, image_height, image_width).get()
  result = [image, scaled_boxes]
  if keypoints is not None:
    scaled_keypoints = keypoint_ops.scale(keypoints, image_height, image_width)
    result.append(scaled_keypoints)
  return tuple(result)


# pylint: disable=g-doc-return-or-yield 

def test_scale(self):
    corners = tf.constant([[0, 0, 100, 200], [50, 120, 100, 140]],
                          dtype=tf.float32)
    boxes = box_list.BoxList(corners)
    boxes.add_field('extra_data', tf.constant([[1], [2]]))

    y_scale = tf.constant(1.0/100)
    x_scale = tf.constant(1.0/200)
    scaled_boxes = box_list_ops.scale(boxes, y_scale, x_scale)
    exp_output = [[0, 0, 1, 1], [0.5, 0.6, 1.0, 0.7]]
    with self.test_session() as sess:
      scaled_corners_out = sess.run(scaled_boxes.get())
      self.assertAllClose(scaled_corners_out, exp_output)
      extra_data_out = sess.run(scaled_boxes.get_field('extra_data'))
      self.assertAllEqual(extra_data_out, [[1], [2]]) 

def scale_boxes_to_pixel_coordinates(image, boxes, keypoints=None):
  """Scales boxes from normalized to pixel coordinates.

  Args:
    image: A 3D float32 tensor of shape [height, width, channels].
    boxes: A 2D float32 tensor of shape [num_boxes, 4] containing the bounding
      boxes in normalized coordinates. Each row is of the form
      [ymin, xmin, ymax, xmax].
    keypoints: (optional) rank 3 float32 tensor with shape
      [num_instances, num_keypoints, 2]. The keypoints are in y-x normalized
      coordinates.

  Returns:
    image: unchanged input image.
    scaled_boxes: a 2D float32 tensor of shape [num_boxes, 4] containing the
      bounding boxes in pixel coordinates.
    scaled_keypoints: a 3D float32 tensor with shape
      [num_instances, num_keypoints, 2] containing the keypoints in pixel
      coordinates.
  """
  boxlist = box_list.BoxList(boxes)
  image_height = tf.shape(image)[0]
  image_width = tf.shape(image)[1]
  scaled_boxes = box_list_ops.scale(boxlist, image_height, image_width).get()
  result = [image, scaled_boxes]
  if keypoints is not None:
    scaled_keypoints = keypoint_ops.scale(keypoints, image_height, image_width)
    result.append(scaled_keypoints)
  return tuple(result)


# TODO(alirezafathi): Investigate if instead the function should return None if
# masks is None.
# pylint: disable=g-doc-return-or-yield 

def test_scale(self):
    corners = tf.constant([[0, 0, 100, 200], [50, 120, 100, 140]],
                          dtype=tf.float32)
    boxes = box_list.BoxList(corners)
    boxes.add_field('extra_data', tf.constant([[1], [2]]))

    y_scale = tf.constant(1.0/100)
    x_scale = tf.constant(1.0/200)
    scaled_boxes = box_list_ops.scale(boxes, y_scale, x_scale)
    exp_output = [[0, 0, 1, 1], [0.5, 0.6, 1.0, 0.7]]
    with self.test_session() as sess:
      scaled_corners_out = sess.run(scaled_boxes.get())
      self.assertAllClose(scaled_corners_out, exp_output)
      extra_data_out = sess.run(scaled_boxes.get_field('extra_data'))
      self.assertAllEqual(extra_data_out, [[1], [2]]) 

def _compute_object_detection_losses(self, input_height, input_width,
                                       prediction_dict, per_pixel_weights):
    """Computes the weighted object detection losses.

    This wrapper function calls the function which computes the losses for
    object detection task and applies corresponding weights to the losses.

    Args:
      input_height: An integer scalar tensor representing input image height.
      input_width: An integer scalar tensor representing input image width.
      prediction_dict: A dictionary holding predicted tensors output by
        "predict" function. See "predict" function for more detailed
        description.
      per_pixel_weights: A float tensor of shape [batch_size,
        out_height * out_width, 1] with 1s in locations where the spatial
        coordinates fall within the height and width in true_image_shapes.

    Returns:
      A dictionary of scalar float tensors representing the weighted losses for
      object detection task:
         BOX_SCALE: the weighted scale (height/width) loss.
         BOX_OFFSET: the weighted object offset loss.
    """
    od_scale_loss, od_offset_loss = self._compute_box_scale_and_offset_loss(
        scale_predictions=prediction_dict[BOX_SCALE],
        offset_predictions=prediction_dict[BOX_OFFSET],
        input_height=input_height,
        input_width=input_width)
    loss_dict = {}
    loss_dict[BOX_SCALE] = (
        self._od_params.scale_loss_weight * od_scale_loss)
    loss_dict[BOX_OFFSET] = (
        self._od_params.offset_loss_weight * od_offset_loss)
    return loss_dict 

def test_scale(self):
    def graph_fn():
      corners = tf.constant([[0, 0, 100, 200], [50, 120, 100, 140]],
                            dtype=tf.float32)
      boxes = box_list.BoxList(corners)
      boxes.add_field('extra_data', tf.constant([[1], [2]]))

      y_scale = tf.constant(1.0/100)
      x_scale = tf.constant(1.0/200)
      scaled_boxes = box_list_ops.scale(boxes, y_scale, x_scale)
      return scaled_boxes.get(), scaled_boxes.get_field('extra_data')
    scaled_corners_out, extra_data_out = self.execute(graph_fn, [])
    exp_output = [[0, 0, 1, 1], [0.5, 0.6, 1.0, 0.7]]
    self.assertAllClose(scaled_corners_out, exp_output)
    self.assertAllEqual(extra_data_out, [[1], [2]]) 

def scale_boxes_to_pixel_coordinates(image, boxes, keypoints=None):
  """Scales boxes from normalized to pixel coordinates.

  Args:
    image: A 3D float32 tensor of shape [height, width, channels].
    boxes: A 2D float32 tensor of shape [num_boxes, 4] containing the bounding
      boxes in normalized coordinates. Each row is of the form
      [ymin, xmin, ymax, xmax].
    keypoints: (optional) rank 3 float32 tensor with shape
      [num_instances, num_keypoints, 2]. The keypoints are in y-x normalized
      coordinates.

  Returns:
    image: unchanged input image.
    scaled_boxes: a 2D float32 tensor of shape [num_boxes, 4] containing the
      bounding boxes in pixel coordinates.
    scaled_keypoints: a 3D float32 tensor with shape
      [num_instances, num_keypoints, 2] containing the keypoints in pixel
      coordinates.
  """
  boxlist = box_list.BoxList(boxes)
  image_height = tf.shape(image)[0]
  image_width = tf.shape(image)[1]
  scaled_boxes = box_list_ops.scale(boxlist, image_height, image_width).get()
  result = [image, scaled_boxes]
  if keypoints is not None:
    scaled_keypoints = keypoint_ops.scale(keypoints, image_height, image_width)
    result.append(scaled_keypoints)
  return tuple(result)


# TODO(alirezafathi): Investigate if instead the function should return None if
# masks is None.
# pylint: disable=g-doc-return-or-yield 

def test_scale(self):
    corners = tf.constant([[0, 0, 100, 200], [50, 120, 100, 140]],
                          dtype=tf.float32)
    boxes = box_list.BoxList(corners)
    boxes.add_field('extra_data', tf.constant([[1], [2]]))

    y_scale = tf.constant(1.0/100)
    x_scale = tf.constant(1.0/200)
    scaled_boxes = box_list_ops.scale(boxes, y_scale, x_scale)
    exp_output = [[0, 0, 1, 1], [0.5, 0.6, 1.0, 0.7]]
    with self.test_session() as sess:
      scaled_corners_out = sess.run(scaled_boxes.get())
      self.assertAllClose(scaled_corners_out, exp_output)
      extra_data_out = sess.run(scaled_boxes.get_field('extra_data'))
      self.assertAllEqual(extra_data_out, [[1], [2]]) 

def scale_boxes_to_pixel_coordinates(image, boxes, keypoints=None):
  """Scales boxes from normalized to pixel coordinates.

  Args:
    image: A 3D float32 tensor of shape [height, width, channels].
    boxes: A 2D float32 tensor of shape [num_boxes, 4] containing the bounding
      boxes in normalized coordinates. Each row is of the form
      [ymin, xmin, ymax, xmax].
    keypoints: (optional) rank 3 float32 tensor with shape
      [num_instances, num_keypoints, 2]. The keypoints are in y-x normalized
      coordinates.

  Returns:
    image: unchanged input image.
    scaled_boxes: a 2D float32 tensor of shape [num_boxes, 4] containing the
      bounding boxes in pixel coordinates.
    scaled_keypoints: a 3D float32 tensor with shape
      [num_instances, num_keypoints, 2] containing the keypoints in pixel
      coordinates.
  """
  boxlist = box_list.BoxList(boxes)
  image_height = tf.shape(image)[0]
  image_width = tf.shape(image)[1]
  scaled_boxes = box_list_ops.scale(boxlist, image_height, image_width).get()
  result = [image, scaled_boxes]
  if keypoints is not None:
    scaled_keypoints = keypoint_ops.scale(keypoints, image_height, image_width)
    result.append(scaled_keypoints)
  return tuple(result)


# pylint: disable=g-doc-return-or-yield 

def test_scale(self):
    corners = tf.constant([[0, 0, 100, 200], [50, 120, 100, 140]],
                          dtype=tf.float32)
    boxes = box_list.BoxList(corners)
    boxes.add_field('extra_data', tf.constant([[1], [2]]))

    y_scale = tf.constant(1.0/100)
    x_scale = tf.constant(1.0/200)
    scaled_boxes = box_list_ops.scale(boxes, y_scale, x_scale)
    exp_output = [[0, 0, 1, 1], [0.5, 0.6, 1.0, 0.7]]
    with self.test_session() as sess:
      scaled_corners_out = sess.run(scaled_boxes.get())
      self.assertAllClose(scaled_corners_out, exp_output)
      extra_data_out = sess.run(scaled_boxes.get_field('extra_data'))
      self.assertAllEqual(extra_data_out, [[1], [2]]) 

def scale_boxes_to_pixel_coordinates(image, boxes, keypoints=None):
  """Scales boxes from normalized to pixel coordinates.

  Args:
    image: A 3D float32 tensor of shape [height, width, channels].
    boxes: A 2D float32 tensor of shape [num_boxes, 4] containing the bounding
      boxes in normalized coordinates. Each row is of the form
      [ymin, xmin, ymax, xmax].
    keypoints: (optional) rank 3 float32 tensor with shape
      [num_instances, num_keypoints, 2]. The keypoints are in y-x normalized
      coordinates.

  Returns:
    image: unchanged input image.
    scaled_boxes: a 2D float32 tensor of shape [num_boxes, 4] containing the
      bounding boxes in pixel coordinates.
    scaled_keypoints: a 3D float32 tensor with shape
      [num_instances, num_keypoints, 2] containing the keypoints in pixel
      coordinates.
  """
  boxlist = box_list.BoxList(boxes)
  image_height = tf.shape(image)[0]
  image_width = tf.shape(image)[1]
  scaled_boxes = box_list_ops.scale(boxlist, image_height, image_width).get()
  result = [image, scaled_boxes]
  if keypoints is not None:
    scaled_keypoints = keypoint_ops.scale(keypoints, image_height, image_width)
    result.append(scaled_keypoints)
  return tuple(result)


# pylint: disable=g-doc-return-or-yield 

def convert_strided_predictions_to_normalized_boxes(boxes, stride,
                                                    true_image_shapes):
  """Converts predictions in the output space to normalized boxes.

  Boxes falling outside the valid image boundary are clipped to be on the
  boundary.

  Args:
    boxes: A tensor of shape [batch_size, num_boxes, 4] holding the raw
     coordinates of boxes in the model's output space.
    stride: The stride in the output space.
    true_image_shapes: A tensor of shape [batch_size, 3] representing the true
      shape of the input not considering padding.

  Returns:
    boxes: A tensor of shape [batch_size, num_boxes, 4] representing the
      coordinates of the normalized boxes.
  """

  def _normalize_boxlist(args):

    boxes, height, width = args
    boxes = box_list_ops.scale(boxes, stride, stride)
    boxes = box_list_ops.to_normalized_coordinates(boxes, height, width)
    boxes = box_list_ops.clip_to_window(boxes, [0., 0., 1., 1.],
                                        filter_nonoverlapping=False)
    return boxes

  box_lists = [box_list.BoxList(boxes) for boxes in tf.unstack(boxes, axis=0)]
  true_heights, true_widths, _ = tf.unstack(true_image_shapes, axis=1)

  true_heights_list = tf.unstack(true_heights, axis=0)
  true_widths_list = tf.unstack(true_widths, axis=0)

  box_lists = list(map(_normalize_boxlist,
                       zip(box_lists, true_heights_list, true_widths_list)))
  boxes = tf.stack([box_list_instance.get() for
                    box_list_instance in box_lists], axis=0)

  return boxes 

def scale_boxes_to_pixel_coordinates(image, boxes, keypoints=None):
    """Scales boxes from normalized to pixel coordinates.

    Args:
      image: A 3D float32 tensor of shape [height, width, channels].
      boxes: A 2D float32 tensor of shape [num_boxes, 4] containing the bounding
        boxes in normalized coordinates. Each row is of the form
        [ymin, xmin, ymax, xmax].
      keypoints: (optional) rank 3 float32 tensor with shape
        [num_instances, num_keypoints, 2]. The keypoints are in y-x normalized
        coordinates.

    Returns:
      image: unchanged input image.
      scaled_boxes: a 2D float32 tensor of shape [num_boxes, 4] containing the
        bounding boxes in pixel coordinates.
      scaled_keypoints: a 3D float32 tensor with shape
        [num_instances, num_keypoints, 2] containing the keypoints in pixel
        coordinates.
    """
    boxlist = box_list.BoxList(boxes)
    image_height = tf.shape(image)[0]
    image_width = tf.shape(image)[1]
    scaled_boxes = box_list_ops.scale(boxlist, image_height, image_width).get()
    result = [image, scaled_boxes]
    if keypoints is not None:
        scaled_keypoints = keypoint_ops.scale(keypoints, image_height, image_width)
        result.append(scaled_keypoints)
    return tuple(result)


# TODO(alirezafathi): Investigate if instead the function should return None if
# masks is None.
# pylint: disable=g-doc-return-or-yield 

def test_scale(self):
        corners = tf.constant([[0, 0, 100, 200], [50, 120, 100, 140]],
                              dtype=tf.float32)
        boxes = box_list.BoxList(corners)
        boxes.add_field('extra_data', tf.constant([[1], [2]]))

        y_scale = tf.constant(1.0 / 100)
        x_scale = tf.constant(1.0 / 200)
        scaled_boxes = box_list_ops.scale(boxes, y_scale, x_scale)
        exp_output = [[0, 0, 1, 1], [0.5, 0.6, 1.0, 0.7]]
        with self.test_session() as sess:
            scaled_corners_out = sess.run(scaled_boxes.get())
            self.assertAllClose(scaled_corners_out, exp_output)
            extra_data_out = sess.run(scaled_boxes.get_field('extra_data'))
            self.assertAllEqual(extra_data_out, [[1], [2]]) 

def test_scale(self):
    corners = tf.constant([[0, 0, 100, 200], [50, 120, 100, 140]],
                          dtype=tf.float32)
    boxes = box_list.BoxList(corners)
    boxes.add_field('extra_data', tf.constant([[1], [2]]))

    y_scale = tf.constant(1.0/100)
    x_scale = tf.constant(1.0/200)
    scaled_boxes = box_list_ops.scale(boxes, y_scale, x_scale)
    exp_output = [[0, 0, 1, 1], [0.5, 0.6, 1.0, 0.7]]
    with self.test_session() as sess:
      scaled_corners_out = sess.run(scaled_boxes.get())
      self.assertAllClose(scaled_corners_out, exp_output)
      extra_data_out = sess.run(scaled_boxes.get_field('extra_data'))
      self.assertAllEqual(extra_data_out, [[1], [2]]) 

def scale_boxes_to_pixel_coordinates(image, boxes, keypoints=None):
  """Scales boxes from normalized to pixel coordinates.

  Args:
    image: A 3D float32 tensor of shape [height, width, channels].
    boxes: A 2D float32 tensor of shape [num_boxes, 4] containing the bounding
      boxes in normalized coordinates. Each row is of the form
      [ymin, xmin, ymax, xmax].
    keypoints: (optional) rank 3 float32 tensor with shape
      [num_instances, num_keypoints, 2]. The keypoints are in y-x normalized
      coordinates.

  Returns:
    image: unchanged input image.
    scaled_boxes: a 2D float32 tensor of shape [num_boxes, 4] containing the
      bounding boxes in pixel coordinates.
    scaled_keypoints: a 3D float32 tensor with shape
      [num_instances, num_keypoints, 2] containing the keypoints in pixel
      coordinates.
  """
  boxlist = box_list.BoxList(boxes)
  image_height = tf.shape(image)[0]
  image_width = tf.shape(image)[1]
  scaled_boxes = box_list_ops.scale(boxlist, image_height, image_width).get()
  result = [image, scaled_boxes]
  if keypoints is not None:
    scaled_keypoints = keypoint_ops.scale(keypoints, image_height, image_width)
    result.append(scaled_keypoints)
  return tuple(result)


# TODO(alirezafathi): Investigate if instead the function should return None if
# masks is None.
# pylint: disable=g-doc-return-or-yield 

def test_scale(self):
    corners = tf.constant([[0, 0, 100, 200], [50, 120, 100, 140]],
                          dtype=tf.float32)
    boxes = box_list.BoxList(corners)
    boxes.add_field('extra_data', tf.constant([[1], [2]]))

    y_scale = tf.constant(1.0/100)
    x_scale = tf.constant(1.0/200)
    scaled_boxes = box_list_ops.scale(boxes, y_scale, x_scale)
    exp_output = [[0, 0, 1, 1], [0.5, 0.6, 1.0, 0.7]]
    with self.test_session() as sess:
      scaled_corners_out = sess.run(scaled_boxes.get())
      self.assertAllClose(scaled_corners_out, exp_output)
      extra_data_out = sess.run(scaled_boxes.get_field('extra_data'))
      self.assertAllEqual(extra_data_out, [[1], [2]]) 

def scale_boxes_to_pixel_coordinates(image, boxes, keypoints=None):
  """Scales boxes from normalized to pixel coordinates.

  Args:
    image: A 3D float32 tensor of shape [height, width, channels].
    boxes: A 2D float32 tensor of shape [num_boxes, 4] containing the bounding
      boxes in normalized coordinates. Each row is of the form
      [ymin, xmin, ymax, xmax].
    keypoints: (optional) rank 3 float32 tensor with shape
      [num_instances, num_keypoints, 2]. The keypoints are in y-x normalized
      coordinates.

  Returns:
    image: unchanged input image.
    scaled_boxes: a 2D float32 tensor of shape [num_boxes, 4] containing the
      bounding boxes in pixel coordinates.
    scaled_keypoints: a 3D float32 tensor with shape
      [num_instances, num_keypoints, 2] containing the keypoints in pixel
      coordinates.
  """
  boxlist = box_list.BoxList(boxes)
  image_height = tf.shape(image)[0]
  image_width = tf.shape(image)[1]
  scaled_boxes = box_list_ops.scale(boxlist, image_height, image_width).get()
  result = [image, scaled_boxes]
  if keypoints is not None:
    scaled_keypoints = keypoint_ops.scale(keypoints, image_height, image_width)
    result.append(scaled_keypoints)
  return tuple(result)


# TODO: Investigate if instead the function should return None if
# masks is None.
# pylint: disable=g-doc-return-or-yield 

def test_scale(self):
    corners = tf.constant([[0, 0, 100, 200], [50, 120, 100, 140]],
                          dtype=tf.float32)
    boxes = box_list.BoxList(corners)
    boxes.add_field('extra_data', tf.constant([[1], [2]]))

    y_scale = tf.constant(1.0/100)
    x_scale = tf.constant(1.0/200)
    scaled_boxes = box_list_ops.scale(boxes, y_scale, x_scale)
    exp_output = [[0, 0, 1, 1], [0.5, 0.6, 1.0, 0.7]]
    with self.test_session() as sess:
      scaled_corners_out = sess.run(scaled_boxes.get())
      self.assertAllClose(scaled_corners_out, exp_output)
      extra_data_out = sess.run(scaled_boxes.get_field('extra_data'))
      self.assertAllEqual(extra_data_out, [[1], [2]]) 

def scale_boxes_to_pixel_coordinates(image, boxes, keypoints=None):
  """Scales boxes from normalized to pixel coordinates.

  Args:
    image: A 3D float32 tensor of shape [height, width, channels].
    boxes: A 2D float32 tensor of shape [num_boxes, 4] containing the bounding
      boxes in normalized coordinates. Each row is of the form
      [ymin, xmin, ymax, xmax].
    keypoints: (optional) rank 3 float32 tensor with shape
      [num_instances, num_keypoints, 2]. The keypoints are in y-x normalized
      coordinates.

  Returns:
    image: unchanged input image.
    scaled_boxes: a 2D float32 tensor of shape [num_boxes, 4] containing the
      bounding boxes in pixel coordinates.
    scaled_keypoints: a 3D float32 tensor with shape
      [num_instances, num_keypoints, 2] containing the keypoints in pixel
      coordinates.
  """
  boxlist = box_list.BoxList(boxes)
  image_height = tf.shape(image)[0]
  image_width = tf.shape(image)[1]
  scaled_boxes = box_list_ops.scale(boxlist, image_height, image_width).get()
  result = [image, scaled_boxes]
  if keypoints is not None:
    scaled_keypoints = keypoint_ops.scale(keypoints, image_height, image_width)
    result.append(scaled_keypoints)
  return tuple(result)


# pylint: disable=g-doc-return-or-yield 

def test_scale(self):
    corners = tf.constant([[0, 0, 100, 200], [50, 120, 100, 140]],
                          dtype=tf.float32)
    boxes = box_list.BoxList(corners)
    boxes.add_field('extra_data', tf.constant([[1], [2]]))

    y_scale = tf.constant(1.0/100)
    x_scale = tf.constant(1.0/200)
    scaled_boxes = box_list_ops.scale(boxes, y_scale, x_scale)
    exp_output = [[0, 0, 1, 1], [0.5, 0.6, 1.0, 0.7]]
    with self.test_session() as sess:
      scaled_corners_out = sess.run(scaled_boxes.get())
      self.assertAllClose(scaled_corners_out, exp_output)
      extra_data_out = sess.run(scaled_boxes.get_field('extra_data'))
      self.assertAllEqual(extra_data_out, [[1], [2]]) 

def scale_boxes_to_pixel_coordinates(image, boxes, keypoints=None):
  """Scales boxes from normalized to pixel coordinates.

  Args:
    image: A 3D float32 tensor of shape [height, width, channels].
    boxes: A 2D float32 tensor of shape [num_boxes, 4] containing the bounding
      boxes in normalized coordinates. Each row is of the form
      [ymin, xmin, ymax, xmax].
    keypoints: (optional) rank 3 float32 tensor with shape
      [num_instances, num_keypoints, 2]. The keypoints are in y-x normalized
      coordinates.

  Returns:
    image: unchanged input image.
    scaled_boxes: a 2D float32 tensor of shape [num_boxes, 4] containing the
      bounding boxes in pixel coordinates.
    scaled_keypoints: a 3D float32 tensor with shape
      [num_instances, num_keypoints, 2] containing the keypoints in pixel
      coordinates.
  """
  boxlist = box_list.BoxList(boxes)
  image_height = tf.shape(image)[0]
  image_width = tf.shape(image)[1]
  scaled_boxes = box_list_ops.scale(boxlist, image_height, image_width).get()
  result = [image, scaled_boxes]
  if keypoints is not None:
    scaled_keypoints = keypoint_ops.scale(keypoints, image_height, image_width)
    result.append(scaled_keypoints)
  return tuple(result)


# pylint: disable=g-doc-return-or-yield 

def test_scale(self):
    corners = tf.constant([[0, 0, 100, 200], [50, 120, 100, 140]],
                          dtype=tf.float32)
    boxes = box_list.BoxList(corners)
    boxes.add_field('extra_data', tf.constant([[1], [2]]))

    y_scale = tf.constant(1.0/100)
    x_scale = tf.constant(1.0/200)
    scaled_boxes = box_list_ops.scale(boxes, y_scale, x_scale)
    exp_output = [[0, 0, 1, 1], [0.5, 0.6, 1.0, 0.7]]
    with self.test_session() as sess:
      scaled_corners_out = sess.run(scaled_boxes.get())
      self.assertAllClose(scaled_corners_out, exp_output)
      extra_data_out = sess.run(scaled_boxes.get_field('extra_data'))
      self.assertAllEqual(extra_data_out, [[1], [2]]) 

def scale_boxes_to_pixel_coordinates(image, boxes, keypoints=None):
  """Scales boxes from normalized to pixel coordinates.

  Args:
    image: A 3D float32 tensor of shape [height, width, channels].
    boxes: A 2D float32 tensor of shape [num_boxes, 4] containing the bounding
      boxes in normalized coordinates. Each row is of the form
      [ymin, xmin, ymax, xmax].
    keypoints: (optional) rank 3 float32 tensor with shape
      [num_instances, num_keypoints, 2]. The keypoints are in y-x normalized
      coordinates.

  Returns:
    image: unchanged input image.
    scaled_boxes: a 2D float32 tensor of shape [num_boxes, 4] containing the
      bounding boxes in pixel coordinates.
    scaled_keypoints: a 3D float32 tensor with shape
      [num_instances, num_keypoints, 2] containing the keypoints in pixel
      coordinates.
  """
  boxlist = box_list.BoxList(boxes)
  image_height = tf.shape(image)[0]
  image_width = tf.shape(image)[1]
  scaled_boxes = box_list_ops.scale(boxlist, image_height, image_width).get()
  result = [image, scaled_boxes]
  if keypoints is not None:
    scaled_keypoints = keypoint_ops.scale(keypoints, image_height, image_width)
    result.append(scaled_keypoints)
  return tuple(result)


# pylint: disable=g-doc-return-or-yield 

def test_scale(self):
    corners = tf.constant([[0, 0, 100, 200], [50, 120, 100, 140]],
                          dtype=tf.float32)
    boxes = box_list.BoxList(corners)
    boxes.add_field('extra_data', tf.constant([[1], [2]]))

    y_scale = tf.constant(1.0/100)
    x_scale = tf.constant(1.0/200)
    scaled_boxes = box_list_ops.scale(boxes, y_scale, x_scale)
    exp_output = [[0, 0, 1, 1], [0.5, 0.6, 1.0, 0.7]]
    with self.test_session() as sess:
      scaled_corners_out = sess.run(scaled_boxes.get())
      self.assertAllClose(scaled_corners_out, exp_output)
      extra_data_out = sess.run(scaled_boxes.get_field('extra_data'))
      self.assertAllEqual(extra_data_out, [[1], [2]]) 

def scale_boxes_to_pixel_coordinates(image, boxes, keypoints=None):
  """Scales boxes from normalized to pixel coordinates.

  Args:
    image: A 3D float32 tensor of shape [height, width, channels].
    boxes: A 2D float32 tensor of shape [num_boxes, 4] containing the bounding
      boxes in normalized coordinates. Each row is of the form
      [ymin, xmin, ymax, xmax].
    keypoints: (optional) rank 3 float32 tensor with shape
      [num_instances, num_keypoints, 2]. The keypoints are in y-x normalized
      coordinates.

  Returns:
    image: unchanged input image.
    scaled_boxes: a 2D float32 tensor of shape [num_boxes, 4] containing the
      bounding boxes in pixel coordinates.
    scaled_keypoints: a 3D float32 tensor with shape
      [num_instances, num_keypoints, 2] containing the keypoints in pixel
      coordinates.
  """
  boxlist = box_list.BoxList(boxes)
  image_height = tf.shape(image)[0]
  image_width = tf.shape(image)[1]
  scaled_boxes = box_list_ops.scale(boxlist, image_height, image_width).get()
  result = [image, scaled_boxes]
  if keypoints is not None:
    scaled_keypoints = keypoint_ops.scale(keypoints, image_height, image_width)
    result.append(scaled_keypoints)
  return tuple(result)


# pylint: disable=g-doc-return-or-yield 

def test_scale(self):
    corners = tf.constant([[0, 0, 100, 200], [50, 120, 100, 140]],
                          dtype=tf.float32)
    boxes = box_list.BoxList(corners)
    boxes.add_field('extra_data', tf.constant([[1], [2]]))

    y_scale = tf.constant(1.0/100)
    x_scale = tf.constant(1.0/200)
    scaled_boxes = box_list_ops.scale(boxes, y_scale, x_scale)
    exp_output = [[0, 0, 1, 1], [0.5, 0.6, 1.0, 0.7]]
    with self.test_session() as sess:
      scaled_corners_out = sess.run(scaled_boxes.get())
      self.assertAllClose(scaled_corners_out, exp_output)
      extra_data_out = sess.run(scaled_boxes.get_field('extra_data'))
      self.assertAllEqual(extra_data_out, [[1], [2]]) 

def scale_boxes_to_pixel_coordinates(image, boxes, keypoints=None):
  """Scales boxes from normalized to pixel coordinates.

  Args:
    image: A 3D float32 tensor of shape [height, width, channels].
    boxes: A 2D float32 tensor of shape [num_boxes, 4] containing the bounding
      boxes in normalized coordinates. Each row is of the form
      [ymin, xmin, ymax, xmax].
    keypoints: (optional) rank 3 float32 tensor with shape
      [num_instances, num_keypoints, 2]. The keypoints are in y-x normalized
      coordinates.

  Returns:
    image: unchanged input image.
    scaled_boxes: a 2D float32 tensor of shape [num_boxes, 4] containing the
      bounding boxes in pixel coordinates.
    scaled_keypoints: a 3D float32 tensor with shape
      [num_instances, num_keypoints, 2] containing the keypoints in pixel
      coordinates.
  """
  boxlist = box_list.BoxList(boxes)
  image_height = tf.shape(image)[0]
  image_width = tf.shape(image)[1]
  scaled_boxes = box_list_ops.scale(boxlist, image_height, image_width).get()
  result = [image, scaled_boxes]
  if keypoints is not None:
    scaled_keypoints = keypoint_ops.scale(keypoints, image_height, image_width)
    result.append(scaled_keypoints)
  return tuple(result)


# TODO(alirezafathi): Investigate if instead the function should return None if
# masks is None.
# pylint: disable=g-doc-return-or-yield