def array_to_base64_png(array):
"""Convert an array into base64-enoded PNG image.
Args:
array: A 2D np.ndarray or nested list of items.
Returns:
A base64-encoded string the image. The image is grayscale if the array is
2D. The image is RGB color if the image is 3D with lsat dimension equal to
3.
"""
# TODO(cais): Deal with 3D case.
# TODO(cais): If there are None values in here, replace them with all NaNs.
array = np.array(array, dtype=np.float32)
assert len(array.shape) == 2
healthy_indices = np.where(np.logical_and(np.logical_not(np.isnan(array)),
np.logical_not(np.isinf(array))))
# TODO(cais): Deal with case in which there is no health elements, i.e., all
# elements are NaN of Inf.
minval = np.min(array[healthy_indices])
maxval = np.max(array[healthy_indices])
# TODO(cais): Deal with the case in which minval == maxval.
scaled = np.array((array - minval) / (maxval - minval) * 255, dtype=np.uint8)
rgb = np.repeat(np.expand_dims(scaled, -1), 3, axis=-1)
image_encoded = base64.b64encode(util.encode_png(rgb))
return image_encoded
def array_to_base64_png(array):
"""Convert an array into base64-enoded PNG image.
Args:
array: A 2D np.ndarray or nested list of items.
Returns:
A base64-encoded string the image. The image is grayscale if the array is
2D. The image is RGB color if the image is 3D with lsat dimension equal to
3.
Raises:
ValueError: If the input `array` is not rank-2, or if the rank-2 `array` is
empty.
"""
# TODO(cais): Deal with 3D case.
# TODO(cais): If there are None values in here, replace them with all NaNs.
array = np.array(array, dtype=np.float32)
if len(array.shape) != 2:
raise ValueError("Expected rank-2 array; received rank-%d array." %
len(array.shape))
if not np.size(array):
raise ValueError("Cannot encode an empty array (size: %s) as image." %
(array.shape, ))
is_infinity = np.isinf(array)
is_positive = array > 0.0
is_positive_infinity = np.logical_and(is_infinity, is_positive)
is_negative_infinity = np.logical_and(is_infinity,
np.logical_not(is_positive))
is_nan = np.isnan(array)
finite_indices = np.where(
np.logical_and(np.logical_not(is_infinity), np.logical_not(is_nan)))
if np.size(finite_indices):
# Finite subset is not empty.
minval = np.min(array[finite_indices])
maxval = np.max(array[finite_indices])
scaled = np.array((array - minval) / (maxval - minval) * 255,
dtype=np.uint8)
rgb = np.repeat(np.expand_dims(scaled, -1),
IMAGE_COLOR_CHANNELS,
axis=-1)
else:
rgb = np.zeros(array.shape + (IMAGE_COLOR_CHANNELS, ), dtype=np.uint8)
# Color-code pixels that correspond to infinities and nans.
rgb[is_positive_infinity] = POSITIVE_INFINITY_RGB
rgb[is_negative_infinity] = NEGATIVE_INFINITY_RGB
rgb[is_nan] = NAN_RGB
image_encoded = base64.b64encode(util.encode_png(rgb))
return image_encoded
def array_to_base64_png(array):
"""Convert an array into base64-enoded PNG image.
Args:
array: A 2D np.ndarray or nested list of items.
Returns:
A base64-encoded string the image. The image is grayscale if the array is
2D. The image is RGB color if the image is 3D with lsat dimension equal to
3.
Raises:
ValueError: If the input `array` is not rank-2, or if the rank-2 `array` is
empty.
"""
# TODO(cais): Deal with 3D case.
# TODO(cais): If there are None values in here, replace them with all NaNs.
array = np.array(array, dtype=np.float32)
if len(array.shape) != 2:
raise ValueError(
"Expected rank-2 array; received rank-%d array." % len(array.shape))
if not np.size(array):
raise ValueError(
"Cannot encode an empty array (size: %s) as image." % (array.shape,))
is_infinity = np.isinf(array)
is_positive = array > 0.0
is_positive_infinity = np.logical_and(is_infinity, is_positive)
is_negative_infinity = np.logical_and(is_infinity,
np.logical_not(is_positive))
is_nan = np.isnan(array)
finite_indices = np.where(np.logical_and(np.logical_not(is_infinity),
np.logical_not(is_nan)))
if np.size(finite_indices):
# Finite subset is not empty.
minval = np.min(array[finite_indices])
maxval = np.max(array[finite_indices])
scaled = np.array((array - minval) / (maxval - minval) * 255,
dtype=np.uint8)
rgb = np.repeat(np.expand_dims(scaled, -1), IMAGE_COLOR_CHANNELS, axis=-1)
else:
rgb = np.zeros(array.shape + (IMAGE_COLOR_CHANNELS,), dtype=np.uint8)
# Color-code pixels that correspond to infinities and nans.
rgb[is_positive_infinity] = POSITIVE_INFINITY_RGB
rgb[is_negative_infinity] = NEGATIVE_INFINITY_RGB
rgb[is_nan] = NAN_RGB
image_encoded = base64.b64encode(util.encode_png(rgb))
return image_encoded
def pb(name, images, max_outputs=3, display_name=None, description=None):
"""Create an image summary protobuf.
This behaves as if you were to create an `op` with the same arguments
(wrapped with constant tensors where appropriate) and then execute
that summary op in a TensorFlow session.
Arguments:
name: A unique name for the generated summary, including any desired
name scopes.
images: An `np.array` representing pixel data with shape
`[k, h, w, c]`, where `k` is the number of images, `w` and `h` are
the width and height of the images, and `c` is the number of
channels, which should be 1, 3, or 4.
max_outputs: Optional `int`. At most this many images will be
emitted. If more than this many images are provided, the first
`max_outputs` many images will be used and the rest silently
discarded.
display_name: Optional name for this summary in TensorBoard, as a
`str`. Defaults to `name`.
description: Optional long-form description for this summary, as a
`str`. Markdown is supported. Defaults to empty.
Returns:
A `tf.Summary` protobuf object.
"""
images = np.array(images).astype(np.uint8)
if images.ndim != 4:
raise ValueError('Shape %r must have rank 4' % (images.shape, ))
limited_images = images[:max_outputs]
encoded_images = [util.encode_png(image) for image in limited_images]
(width, height) = (images.shape[2], images.shape[1])
content = [str(width), str(height)] + encoded_images
tensor = tf.make_tensor_proto(content, dtype=tf.string)
if display_name is None:
display_name = name
summary_metadata = metadata.create_summary_metadata(
display_name=display_name, description=description)
summary = tf.Summary()
summary.value.add(tag='%s/image_summary' % name,
metadata=summary_metadata,
tensor=tensor)
return summary
def bench(image, thread_count):
"""Encode `image` to PNG on `thread_count` threads in parallel.
Returns:
A `float` representing number of seconds that it takes all threads
to finish encoding `image`.
"""
threads = [threading.Thread(target=lambda: util.encode_png(image))
for _ in xrange(thread_count)]
start_time = datetime.datetime.now()
for thread in threads:
thread.start()
for thread in threads:
thread.join()
end_time = datetime.datetime.now()
delta = (end_time - start_time).total_seconds()
return delta
def pb(name, images, max_outputs=3, display_name=None, description=None):
"""Create an image summary protobuf.
This behaves as if you were to create an `op` with the same arguments
(wrapped with constant tensors where appropriate) and then execute
that summary op in a TensorFlow session.
Arguments:
name: A unique name for the generated summary, including any desired
name scopes.
images: An `np.array` representing pixel data with shape
`[k, w, h, c]`, where `k` is the number of images, `w` and `h` are
the width and height of the images, and `c` is the number of
channels, which should be 1, 3, or 4.
max_outputs: Optional `int`. At most this many images will be
emitted. If more than this many images are provided, the first
`max_outputs` many images will be used and the rest silently
discarded.
display_name: Optional name for this summary in TensorBoard, as a
`str`. Defaults to `name`.
description: Optional long-form description for this summary, as a
`str`. Markdown is supported. Defaults to empty.
Returns:
A `tf.Summary` protobuf object.
"""
images = np.array(images).astype(np.uint8)
if images.ndim != 4:
raise ValueError('Shape %r must have rank 4' % (images.shape, ))
limited_images = images[:max_outputs]
encoded_images = [util.encode_png(image) for image in limited_images]
(width, height) = (images.shape[1], images.shape[2])
content = [str(width), str(height)] + encoded_images
tensor = tf.make_tensor_proto(content, dtype=tf.string)
if display_name is None:
display_name = name
summary_metadata = metadata.create_summary_metadata(
display_name=display_name, description=description)
summary = tf.Summary()
summary.value.add(tag='%s/image_summary' % name,
metadata=summary_metadata,
tensor=tensor)
return summary
def bench(image, thread_count):
"""Encode `image` to PNG on `thread_count` threads in parallel.
Returns:
A `float` representing number of seconds that it takes all threads
to finish encoding `image`.
"""
threads = [
threading.Thread(target=lambda: util.encode_png(image))
for _ in xrange(thread_count)
]
start_time = datetime.datetime.now()
for thread in threads:
thread.start()
for thread in threads:
thread.join()
end_time = datetime.datetime.now()
delta = (end_time - start_time).total_seconds()
return delta
def _frame_generator(self):
while True:
last_duration = 0
if self.FPS == 0:
continue
else:
time.sleep(max(0, 1/(self.FPS) - last_duration))
start_time = time.time()
array = self._fetch_current_frame()
image_bytes = util.encode_png(array)
frame_text = b'--frame\r\n'
content_type = b'Content-Type: image/png\r\n\r\n'
response_content = frame_text + content_type + image_bytes + b'\r\n\r\n'
last_duration = time.time() - start_time
yield response_content
def write_image(array, filename):
with tf.gfile.Open(filename, 'w') as image_file:
image_file.write(util.encode_png(array))
def write_image(array, filename):
with tf.gfile.Open(filename, 'w') as image_file:
image_file.write(util.encode_png(array))
