def testExisting(self):
# Read a real jpeg and verify shape
path = ('tensorflow/core/lib/jpeg/testdata/' 'jpeg_merge_test1.jpg')
with self.test_session() as sess:
jpeg0 = io_ops.read_file(path)
image0 = image_ops.decode_jpeg(jpeg0)
image1 = image_ops.decode_jpeg(image_ops.encode_jpeg(image0))
jpeg0, image0, image1 = sess.run([jpeg0, image0, image1])
self.assertEqual(len(jpeg0), 3771)
self.assertEqual(image0.shape, (256, 128, 3))
self.assertLess(self.averageError(image0, image1), 0.8)
def testExisting(self):
# Read a real jpeg and verify shape
path = "tensorflow/core/lib/jpeg/testdata/" "jpeg_merge_test1.jpg"
with self.test_session() as sess:
jpeg0 = io_ops.read_file(path)
image0 = image_ops.decode_jpeg(jpeg0)
image1 = image_ops.decode_jpeg(image_ops.encode_jpeg(image0))
jpeg0, image0, image1 = sess.run([jpeg0, image0, image1])
self.assertEqual(len(jpeg0), 3771)
self.assertEqual(image0.shape, (256, 128, 3))
self.assertLess(self.averageError(image0, image1), 0.8)
def _evalDecodeJpeg(self, image_name, parallelism, num_iters, tile=None):
"""Evaluate DecodeJpegOp for the given image.
TODO(tanmingxing): add decoding+cropping as well.
Args:
image_name: a string of image file name (without suffix).
parallelism: the number of concurrent decode_jpeg ops to be run.
num_iters: number of iterations for evaluation.
tile: if not None, tile the image to composite a larger fake image.
Returns:
The duration of the run in seconds.
"""
ops.reset_default_graph()
image_file_path = os.path.join(prefix_path, image_name)
if tile is None:
image_content = variable_scope.get_variable(
'image_%s' % image_name,
initializer=io_ops.read_file(image_file_path))
else:
single_image = image_ops.decode_jpeg(
io_ops.read_file(image_file_path),
channels=3,
name='single_image')
# Tile the image to composite a new larger image.
tiled_image = array_ops.tile(single_image, tile)
image_content = variable_scope.get_variable(
'tiled_image_%s' % image_name,
initializer=image_ops.encode_jpeg(tiled_image))
with session.Session() as sess:
sess.run(variables.global_variables_initializer())
images = []
for i in xrange(parallelism):
images.append(
image_ops.decode_jpeg(image_content,
channels=3,
name='image_%d' % (i)))
r = control_flow_ops.group(*images)
for _ in xrange(3):
# Skip warm up time.
sess.run(r)
start_time = time.time()
for _ in xrange(num_iters):
sess.run(r)
return time.time() - start_time
def _evalDecodeJpeg(self, image_name, parallelism, num_iters, tile=None):
"""Evaluate DecodeJpegOp for the given image.
TODO(tanmingxing): add decoding+cropping as well.
Args:
image_name: a string of image file name (without suffix).
parallelism: the number of concurrent decode_jpeg ops to be run.
num_iters: number of iterations for evaluation.
tile: if not None, tile the image to composite a larger fake image.
Returns:
The duration of the run in seconds.
"""
ops.reset_default_graph()
image_file_path = os.path.join(prefix_path, image_name)
if tile is None:
image_content = variable_scope.get_variable(
'image_%s' % image_name,
initializer=io_ops.read_file(image_file_path))
else:
single_image = image_ops.decode_jpeg(
io_ops.read_file(image_file_path), channels=3, name='single_image')
# Tile the image to composite a new larger image.
tiled_image = array_ops.tile(single_image, tile)
image_content = variable_scope.get_variable(
'tiled_image_%s' % image_name,
initializer=image_ops.encode_jpeg(tiled_image))
with session.Session() as sess:
sess.run(variables.global_variables_initializer())
images = []
for i in xrange(parallelism):
images.append(
image_ops.decode_jpeg(
image_content, channels=3, name='image_%d' % (i)))
r = control_flow_ops.group(*images)
for _ in xrange(3):
# Skip warm up time.
sess.run(r)
start_time = time.time()
for _ in xrange(num_iters):
sess.run(r)
return time.time() - start_time
def testCmyk(self):
# Confirm that CMYK reads in as RGB
base = "tensorflow/core/lib/jpeg/testdata"
rgb_path = os.path.join(base, "jpeg_merge_test1.jpg")
cmyk_path = os.path.join(base, "jpeg_merge_test1_cmyk.jpg")
shape = 256, 128, 3
for channels in 3, 0:
with self.test_session() as sess:
rgb = image_ops.decode_jpeg(io_ops.read_file(rgb_path), channels=channels)
cmyk = image_ops.decode_jpeg(io_ops.read_file(cmyk_path), channels=channels)
rgb, cmyk = sess.run([rgb, cmyk])
self.assertEqual(rgb.shape, shape)
self.assertEqual(cmyk.shape, shape)
error = self.averageError(rgb, cmyk)
self.assertLess(error, 4)
def decode_jpeg():
"""Decodes a jpeg image with specified '_dct_method'."""
return math_ops.cast(
image_ops.decode_jpeg(image_buffer,
channels=self._channels,
dct_method=self._dct_method),
self._dtype)
def testShape(self):
with self.test_session() as sess:
jpeg = constant_op.constant('nonsense')
for channels in 0, 1, 3:
image = image_ops.decode_jpeg(jpeg, channels=channels)
self.assertEqual(image.get_shape().as_list(),
[None, None, channels or None])
def decode_jpeg():
"""Decodes a jpeg image with specified '_dct_method'."""
return math_ops.cast(
image_ops.decode_jpeg(
image_buffer,
channels=self._channels,
dct_method=self._dct_method), self._dtype)
def decode(serialized_example):
features = tf.parse_single_example(
serialized_example,
# Defaults are not specified since both keys are required.
features={
'image/name': tf.FixedLenFeature([], tf.string),
'image/height': tf.FixedLenFeature([1], tf.int64),
'image/width': tf.FixedLenFeature([1], tf.int64),
'image/channels': tf.FixedLenFeature([1], tf.int64),
'image/shape': tf.FixedLenFeature([3], tf.int64),
'image/format': tf.FixedLenFeature([], tf.string),
'image/encoded': tf.FixedLenFeature([], tf.string),
'label/segmentation/format': tf.FixedLenFeature([], tf.string),
'label/segmentation/encoded': tf.FixedLenFeature([], tf.string),
'label/object/bbox/xmin': tf.VarLenFeature(tf.float32),
'label/object/bbox/ymin': tf.VarLenFeature(tf.float32),
'label/object/bbox/xmax': tf.VarLenFeature(tf.float32),
'label/object/bbox/ymax': tf.VarLenFeature(tf.float32),
'label/object/bbox/label': tf.VarLenFeature(tf.int64),
'label/object/bbox/difficult': tf.VarLenFeature(tf.int64),
'label/object/bbox/truncated': tf.VarLenFeature(tf.int64)
})
features['image/encoded'] = image_ops.decode_jpeg(
features['image/encoded'], channels=3)
features['label/segmentation/encoded'] = image_ops.decode_png(
features['label/segmentation/encoded'], channels=1)
return features
def decode(serialized_example):
features = tf.parse_single_example(
serialized_example,
# Defaults are not specified since both keys are required.
features={
'image/name': tf.FixedLenFeature([], tf.string),
'image/height': tf.FixedLenFeature([], tf.int64),
'image/width': tf.FixedLenFeature([], tf.int64),
'image/channels': tf.FixedLenFeature([], tf.int64),
'image/format': tf.FixedLenFeature([], tf.string),
'image/encoded': tf.FixedLenFeature([], tf.string),
'label/num_classes': tf.FixedLenFeature([], tf.int64),
'label/masks': tf.FixedLenFeature([], tf.string),
'label/bboxes': tf.FixedLenFeature([], tf.string),
'label/classes': tf.FixedLenFeature([], tf.string)
}
)
features['image/encoded'] = image_ops.decode_jpeg(features['image/encoded'], channels=3)
features['label/masks'] = tf.decode_raw(features['label/masks'], tf.float64)
features['label/bboxes'] = tf.decode_raw(features['label/bboxes'], tf.float64)
features['label/classes'] = tf.decode_raw(features['label/classes'], tf.float64)
ih = tf.cast(features['image/height'], tf.int32)
iw = tf.cast(features['image/width'], tf.int32)
num_classes = tf.cast(features['label/num_classes'], tf.int32)
features['label/masks'] = tf.cast(tf.reshape(features['label/masks'], [ih, iw, num_classes]), tf.int64)
features['label/bboxes'] = tf.cast(tf.reshape(features['label/bboxes'], [num_classes, 4]), tf.float64)
features['label/classes'] = tf.cast(tf.reshape(features['label/classes'], [num_classes, 1]), tf.int64)
return features
def testShape(self):
with self.test_session() as sess:
jpeg = constant_op.constant('nonsense')
for channels in 0, 1, 3:
image = image_ops.decode_jpeg(jpeg, channels=channels)
self.assertEqual(image.get_shape().as_list(),
[None, None, channels or None])
def testCmyk(self):
# Confirm that CMYK reads in as RGB
base = 'tensorflow/core/lib/jpeg/testdata'
rgb_path = os.path.join(base, 'jpeg_merge_test1.jpg')
cmyk_path = os.path.join(base, 'jpeg_merge_test1_cmyk.jpg')
shape = 256, 128, 3
for channels in 3, 0:
with self.test_session() as sess:
rgb = image_ops.decode_jpeg(io_ops.read_file(rgb_path),
channels=channels)
cmyk = image_ops.decode_jpeg(io_ops.read_file(cmyk_path),
channels=channels)
rgb, cmyk = sess.run([rgb, cmyk])
self.assertEqual(rgb.shape, shape)
self.assertEqual(cmyk.shape, shape)
error = self.averageError(rgb, cmyk)
self.assertLess(error, 4)
def testSynthetic(self):
with self.test_session() as sess:
# Encode it, then decode it, then encode it
image0 = constant_op.constant(_SimpleColorRamp())
jpeg0 = image_ops.encode_jpeg(image0)
image1 = image_ops.decode_jpeg(jpeg0)
image2 = image_ops.decode_jpeg(image_ops.encode_jpeg(image1))
jpeg0, image0, image1, image2 = sess.run([jpeg0, image0, image1, image2])
# The decoded-encoded image should be similar to the input
self.assertLess(self.averageError(image0, image1), 0.6)
# We should be very close to a fixpoint
self.assertLess(self.averageError(image1, image2), 0.02)
# Smooth ramps compress well (input size is 153600)
self.assertGreaterEqual(len(jpeg0), 5000)
self.assertLessEqual(len(jpeg0), 6000)
def testSynthetic(self):
with self.test_session() as sess:
# Encode it, then decode it, then encode it
image0 = constant_op.constant(_SimpleColorRamp())
jpeg0 = image_ops.encode_jpeg(image0)
image1 = image_ops.decode_jpeg(jpeg0)
image2 = image_ops.decode_jpeg(image_ops.encode_jpeg(image1))
jpeg0, image0, image1, image2 = sess.run([jpeg0, image0, image1, image2])
# The decoded-encoded image should be similar to the input
self.assertLess(self.averageError(image0, image1), 0.6)
# We should be very close to a fixpoint
self.assertLess(self.averageError(image1, image2), 0.02)
# Smooth ramps compress well (input size is 153600)
self.assertGreaterEqual(len(jpeg0), 5000)
self.assertLessEqual(len(jpeg0), 6000)
def testJpeg(self):
# Read a real jpeg and verify shape
path = os.path.join(prefix_path, "jpeg", "testdata", "jpeg_merge_test1.jpg")
with self.test_session(use_gpu=True) as sess:
jpeg0 = io_ops.read_file(path)
image0 = image_ops.decode_image(jpeg0)
image1 = image_ops.decode_jpeg(jpeg0)
jpeg0, image0, image1 = sess.run([jpeg0, image0, image1])
self.assertEqual(len(jpeg0), 3771)
self.assertEqual(image0.shape, (256, 128, 3))
self.assertAllEqual(image0, image1)
def testJpeg(self):
# Read a real jpeg and verify shape
path = os.path.join(prefix_path, "jpeg", "testdata", "jpeg_merge_test1.jpg")
with self.test_session(use_gpu=True) as sess:
jpeg0 = io_ops.read_file(path)
image0 = image_ops.decode_image(jpeg0)
image1 = image_ops.decode_jpeg(jpeg0)
jpeg0, image0, image1 = sess.run([jpeg0, image0, image1])
self.assertEqual(len(jpeg0), 3771)
self.assertEqual(image0.shape, (256, 128, 3))
self.assertAllEqual(image0, image1)
def extract(features):
image = features['image/encoded']
image = image_ops.decode_jpeg(image, channels=3)
label = features['image/class/label']
ymin = tf.sparse_tensor_to_dense(features['image/object/bbox/ymin'])
xmin = tf.sparse_tensor_to_dense(features['image/object/bbox/xmin'])
ymax = tf.sparse_tensor_to_dense(features['image/object/bbox/ymax'])
xmax = tf.sparse_tensor_to_dense(features['image/object/bbox/xmax'])
bboxes = tf.stack([ymin, xmin, ymax, xmax], axis=-1)
name = features['image/filename']
return name, image, label, bboxes
def testJpeg(self):
# Read a real jpeg and verify shape
path = os.path.join(prefix_path, "jpeg", "testdata", "jpeg_merge_test1.jpg")
with self.session():
jpeg0 = io_ops.read_file(path)
image0 = image_ops.decode_image(jpeg0)
image1 = image_ops.decode_jpeg(jpeg0)
jpeg0, image0, image1 = self.evaluate([jpeg0, image0, image1])
self.assertEqual(len(jpeg0), 3771)
self.assertEqual(image0.shape, (256, 128, 3))
self.assertAllEqual(image0, image1)
with self.assertRaises(errors_impl.InvalidArgumentError):
bad_channels = image_ops.decode_image(jpeg0, channels=4)
self.evaluate(bad_channels)
def testJpeg(self):
# Read a real jpeg and verify shape
path = os.path.join(prefix_path, "jpeg", "testdata", "jpeg_merge_test1.jpg")
with self.session(use_gpu=True) as sess:
jpeg0 = io_ops.read_file(path)
image0 = image_ops.decode_image(jpeg0)
image1 = image_ops.decode_jpeg(jpeg0)
jpeg0, image0, image1 = self.evaluate([jpeg0, image0, image1])
self.assertEqual(len(jpeg0), 3771)
self.assertEqual(image0.shape, (256, 128, 3))
self.assertAllEqual(image0, image1)
bad_channels = image_ops.decode_image(jpeg0, channels=4)
with self.assertRaises(errors_impl.InvalidArgumentError):
self.evaluate(bad_channels)
def decode(serialized_example):
features = tf.parse_single_example(
serialized_example,
# Defaults are not specified since both keys are required.
features={
'image/name': tf.FixedLenFeature([], tf.string),
'image/height': tf.FixedLenFeature([1], tf.int64),
'image/width': tf.FixedLenFeature([1], tf.int64),
'image/channels': tf.FixedLenFeature([1], tf.int64),
'image/shape': tf.FixedLenFeature([3], tf.int64),
'image/format': tf.FixedLenFeature([], tf.string),
'image/encoded': tf.FixedLenFeature([], tf.string),
'label/format': tf.FixedLenFeature([], tf.string),
'label/encoded': tf.FixedLenFeature([], tf.string)
})
features['image/encoded'] = image_ops.decode_jpeg(features['image/encoded'], channels=3)
features['label/encoded'] = image_ops.decode_png(features['label/encoded'], channels=1)
# image.set_shape((mnist.IMAGE_PIXELS))
return features
def decode(serialized_example):
features = tf.parse_single_example(
serialized_example,
# Defaults are not specified since both keys are required.
features={
'image/name': tf.FixedLenFeature([], tf.string),
'image/encoded': tf.FixedLenFeature([], tf.string),
'image/height': tf.FixedLenFeature([], tf.int64),
'image/width': tf.FixedLenFeature([], tf.int64),
'label/encoded': tf.FixedLenFeature([], tf.string)
})
features['image/encoded'] = image_ops.decode_jpeg(
features['image/encoded'], channels=3)
features['label/encoded'] = tf.decode_raw(features['label/encoded'],
tf.float64)
ih = tf.cast(features['image/height'], tf.int32)
iw = tf.cast(features['image/width'], tf.int32)
features['label/encoded'] = tf.reshape(features['label/encoded'],
[ih, iw, 1])
features['label/encoded'] = tf.cast(features['label/encoded'], tf.int64)
# image.set_shape((mnist.IMAGE_PIXELS))
return features
def decode_image(image_buffer):
return image_ops.decode_jpeg(image_buffer, 3)
def decode_jpeg():
return tf.cast(image_ops.decode_jpeg(image_buffer, channels=3),
tf.uint8)
def decode_jpg():
return image_ops.decode_jpeg(image_buffer, self._channels)
def decode_jpg():
if self._dtype != dtypes.uint8:
raise ValueError(
'jpeg decoder can only be used to decode to tf.uint8 but %s was '
'requested for a jpeg image.' % self._dtype)
return image_ops.decode_jpeg(image_buffer, self._channels)
def decode_jpg():
if self._dtype != dtypes.uint8:
raise ValueError(
'jpeg decoder can only be used to decode to tf.uint8 but %s was '
'requested for a jpeg image.' % self._dtype)
return image_ops.decode_jpeg(image_buffer, self._channels)
def _evalDecodeJpeg(self,
image_name,
parallelism,
num_iters,
crop_during_decode=None,
crop_window=None,
tile=None):
"""Evaluate DecodeJpegOp for the given image.
TODO(tanmingxing): add decoding+cropping as well.
Args:
image_name: a string of image file name (without suffix).
parallelism: the number of concurrent decode_jpeg ops to be run.
num_iters: number of iterations for evaluation.
crop_during_decode: If true, use fused DecodeAndCropJpeg instead of
separate decode and crop ops. It is ignored if crop_window is None.
crop_window: if not None, crop the decoded image. Depending on
crop_during_decode, cropping could happen during or after decoding.
tile: if not None, tile the image to composite a larger fake image.
Returns:
The duration of the run in seconds.
"""
ops.reset_default_graph()
image_file_path = os.path.join(prefix_path, image_name)
if tile is None:
image_content = variable_scope.get_variable(
'image_%s' % image_name,
initializer=io_ops.read_file(image_file_path))
else:
single_image = image_ops.decode_jpeg(
io_ops.read_file(image_file_path), channels=3, name='single_image')
# Tile the image to composite a new larger image.
tiled_image = array_ops.tile(single_image, tile)
image_content = variable_scope.get_variable(
'tiled_image_%s' % image_name,
initializer=image_ops.encode_jpeg(tiled_image))
with session.Session() as sess:
sess.run(variables.global_variables_initializer())
images = []
for _ in xrange(parallelism):
if crop_window is None:
# No crop.
image = image_ops.decode_jpeg(image_content, channels=3)
elif crop_during_decode:
# combined decode and crop.
image = image_ops.decode_and_crop_jpeg(
image_content, crop_window, channels=3)
else:
# separate decode and crop.
image = image_ops.decode_jpeg(image_content, channels=3)
image = image_ops.crop_to_bounding_box(
image,
offset_height=crop_window[0],
offset_width=crop_window[1],
target_height=crop_window[2],
target_width=crop_window[3])
images.append(image)
r = control_flow_ops.group(*images)
for _ in xrange(3):
# Skip warm up time.
sess.run(r)
start_time = time.time()
for _ in xrange(num_iters):
sess.run(r)
return time.time() - start_time
def _evalDecodeJpeg(self,
image_name,
parallelism,
num_iters,
crop_during_decode=None,
crop_window=None,
tile=None):
"""Evaluate DecodeJpegOp for the given image.
TODO(tanmingxing): add decoding+cropping as well.
Args:
image_name: a string of image file name (without suffix).
parallelism: the number of concurrent decode_jpeg ops to be run.
num_iters: number of iterations for evaluation.
crop_during_decode: If true, use fused DecodeAndCropJpeg instead of
separate decode and crop ops. It is ignored if crop_window is None.
crop_window: if not None, crop the decoded image. Depending on
crop_during_decode, cropping could happen during or after decoding.
tile: if not None, tile the image to composite a larger fake image.
Returns:
The duration of the run in seconds.
"""
ops.reset_default_graph()
image_file_path = resource_loader.get_path_to_datafile(
os.path.join('core', 'lib', 'jpeg', 'testdata', image_name))
if tile is None:
image_content = variable_scope.get_variable(
'image_%s' % image_name,
initializer=io_ops.read_file(image_file_path))
else:
single_image = image_ops.decode_jpeg(
io_ops.read_file(image_file_path),
channels=3,
name='single_image')
# Tile the image to composite a new larger image.
tiled_image = array_ops.tile(single_image, tile)
image_content = variable_scope.get_variable(
'tiled_image_%s' % image_name,
initializer=image_ops.encode_jpeg(tiled_image))
with session.Session() as sess:
self.evaluate(variables.global_variables_initializer())
images = []
for _ in xrange(parallelism):
if crop_window is None:
# No crop.
image = image_ops.decode_jpeg(image_content, channels=3)
elif crop_during_decode:
# combined decode and crop.
image = image_ops.decode_and_crop_jpeg(image_content,
crop_window,
channels=3)
else:
# separate decode and crop.
image = image_ops.decode_jpeg(image_content, channels=3)
image = image_ops.crop_to_bounding_box(
image,
offset_height=crop_window[0],
offset_width=crop_window[1],
target_height=crop_window[2],
target_width=crop_window[3])
images.append(image)
r = control_flow_ops.group(*images)
for _ in xrange(3):
# Skip warm up time.
self.evaluate(r)
start_time = time.time()
for _ in xrange(num_iters):
self.evaluate(r)
end_time = time.time()
return end_time - start_time
def DecodeJpg():
return image_ops.decode_jpeg(image_buffer, 3)
def DecodeJpg(): return image_ops.decode_jpeg(image_buffer, 3)
def decode_jpg(): return image_ops.decode_jpeg(image_buffer, self._channels)
def decode_jpeg():
return image_ops.decode_jpeg(
image_buffer, channels=num_channels, dct_method=self._dct_method)
