def build(self):
import imageio
pixels = imageio.imread(self.filepath)
pixels = channels_to_front(pixels)
transparent_types = {'.png'}
filepath = Path(self.filepath)
if pixels.shape[0] == 4 and filepath.suffix in transparent_types:
# If normalise is False, then we return the alpha as an extra
# channel, which can be useful if the alpha channel has semantic
# meanings!
if self.normalise:
p = normalize_pixels_range(pixels[:3])
return MaskedImage(p,
mask=pixels[-1].astype(np.bool),
copy=False)
else:
return Image(pixels, copy=False)
# Assumed not to have an Alpha channel
if self.normalise:
return Image(normalize_pixels_range(pixels), copy=False)
else:
return Image(pixels, copy=False)
def _pil_to_numpy(pil_image, normalise, convert=None):
p = pil_image.convert(convert) if convert else pil_image
p = channels_to_front(p)
if normalise:
return normalize_pixels_range(p)
else:
return p
def vector_128_dsift(x, dtype=np.float32):
r"""
Computes a SIFT feature vector from a square patch (or image). Patch
**must** be square and the output vector will *always* be a ``(128,)``
vector. Please see :func:`dsift` for more information.
Parameters
----------
x : :map:`Image` or subclass or ``(C, Y, Y)`` `ndarray`
Either the image object itself or an array with the pixels. The first
dimension is interpreted as channels. Must be square i.e.
``height == width``.
dtype : ``np.dtype``, optional
The dtype of the returned vector.
Raises
------
ValueError
Only square images are supported.
"""
if not isinstance(x, np.ndarray):
x = x.pixels
if x.shape[-1] != x.shape[-2]:
raise ValueError('This feature only works with square images '
'i.e. width == height')
patch_shape = x.shape[-1]
n_bins = 4
c_size = patch_shape // n_bins
x = normalize_pixels_range(x, error_on_unknown_type=False)
return dsift(x,
window_step_horizontal=patch_shape,
window_step_vertical=patch_shape,
num_bins_horizontal=n_bins, num_bins_vertical=n_bins,
cell_size_horizontal=c_size, cell_size_vertical=c_size,
num_or_bins=8, fast=True).astype(dtype)
def vector_128_dsift(x, dtype=np.float32):
r"""
Computes a SIFT feature vector from a square patch (or image). Patch
**must** be square and the output vector will *always* be a ``(128,)``
vector. Please see :func:`dsift` for more information.
Parameters
----------
x : :map:`Image` or subclass or ``(C, Y, Y)`` `ndarray`
Either the image object itself or an array with the pixels. The first
dimension is interpreted as channels. Must be square i.e.
``height == width``.
dtype : ``np.dtype``, optional
The dtype of the returned vector.
Raises
------
ValueError
Only square images are supported.
"""
if not isinstance(x, np.ndarray):
x = x.pixels
if x.shape[-1] != x.shape[-2]:
raise ValueError('This feature only works with square images '
'i.e. width == height')
patch_shape = x.shape[-1]
n_bins = 4
c_size = patch_shape // n_bins
x = normalize_pixels_range(x, error_on_unknown_type=False)
return dsift(x,
window_step_horizontal=patch_shape,
window_step_vertical=patch_shape,
num_bins_horizontal=n_bins, num_bins_vertical=n_bins,
cell_size_horizontal=c_size, cell_size_vertical=c_size,
num_or_bins=8, fast=True).astype(dtype)
def _pil_to_numpy(pil_image, normalize, convert=None):
p = pil_image.convert(convert) if convert else pil_image
p = channels_to_front(p)
if normalize:
return normalize_pixels_range(p)
else:
return p
def _pil_to_numpy(pil_image, normalize, convert=None):
p = pil_image.convert(convert) if convert else pil_image
p = np.asarray(p)
if normalize:
return normalize_pixels_range(p)
else:
return p
def _pil_to_numpy(pil_image, normalize, convert=None):
p = pil_image.convert(convert) if convert else pil_image
p = np.asarray(p)
if normalize:
return normalize_pixels_range(p)
else:
return p
def import_mesh(path):
if path.suffix == '.pkl' or path.suffix == '.gz':
mesh = import_pickle(path)
else:
mesh = m3io.import_mesh(path)
if hasattr(mesh, "texture") and mesh.texture.pixels.dtype != np.float64:
mesh.texture.pixels = normalize_pixels_range(mesh.texture.pixels)
return mesh
def _imageio_to_menpo(imio_reader, normalise, index):
pixels = imio_reader.get_data(index)
pixels = channels_to_front(pixels)
if normalise:
return Image(normalize_pixels_range(pixels), copy=False)
else:
return Image(pixels, copy=False)
def imageio_importer(filepath, asset=None, normalise=True, **kwargs):
r"""
Imports images using the imageio library - which is actually fairly similar
to our importing logic - but contains the necessary plugins to import lots
of interesting image types like RAW images.
Parameters
----------
filepath : `Path`
Absolute filepath of the image.
asset : `object`, optional
An optional asset that may help with loading. This is unused for this
implementation.
normalise : `bool`, optional
If ``True``, normalise between 0.0 and 1.0 and convert to float. If
``False`` just return whatever imageio imports.
\**kwargs : `dict`, optional
Any other keyword arguments.
Returns
-------
image : :map:`Image` or subclass
The imported image.
"""
import imageio
pixels = imageio.imread(str(filepath))
pixels = channels_to_front(pixels)
transparent_types = {'.png'}
if pixels.shape[0] == 4 and filepath.suffix in transparent_types:
# If normalise is False, then we return the alpha as an extra
# channel, which can be useful if the alpha channel has semantic
# meanings!
if normalise:
p = normalize_pixels_range(pixels[:3])
return MaskedImage(p, mask=pixels[-1].astype(np.bool),
copy=False)
else:
return Image(pixels, copy=False)
# Assumed not to have an Alpha channel
if normalise:
return Image(normalize_pixels_range(pixels), copy=False)
else:
return Image(pixels, copy=False)
def _imageio_to_menpo(imio_reader, normalise, index):
pixels = imio_reader.get_data(index)
pixels = channels_to_front(pixels)
if normalise:
return Image(normalize_pixels_range(pixels), copy=False)
else:
return Image(pixels, copy=False)
def imageio_importer(filepath, asset=None, normalize=True, **kwargs):
r"""
Imports images using the imageio library - which is actually fairly similar
to our importing logic - but contains the necessary plugins to import lots
of interesting image types like RAW images.
Parameters
----------
filepath : `Path`
Absolute filepath of the image.
asset : `object`, optional
An optional asset that may help with loading. This is unused for this
implementation.
normalize : `bool`, optional
If ``True``, normalize between 0.0 and 1.0 and convert to float. If
``False`` just return whatever imageio imports.
\**kwargs : `dict`, optional
Any other keyword arguments.
Returns
-------
image : :map:`Image` or subclass
The imported image.
"""
import imageio
pixels = imageio.imread(str(filepath))
pixels = channels_to_front(pixels)
transparent_types = {'.png'}
if pixels.shape[0] == 4 and filepath.suffix in transparent_types:
# If normalize is False, then we return the alpha as an extra
# channel, which can be useful if the alpha channel has semantic
# meanings!
if normalize:
p = normalize_pixels_range(pixels[:3])
return MaskedImage(p, mask=pixels[-1].astype(np.bool),
copy=False)
else:
return Image(pixels, copy=False)
# Assumed not to have an Alpha channel
if normalize:
return Image(normalize_pixels_range(pixels), copy=False)
else:
return Image(pixels, copy=False)
def imageio_to_menpo(imio_reader, index):
pixels = imio_reader.get_data(index)
pixels = channels_to_front(pixels)
if pixels.shape[0] == 4:
# If normalize is False, then we return the alpha as an extra
# channel, which can be useful if the alpha channel has semantic
# meanings!
if normalize:
p = normalize_pixels_range(pixels[:3])
return MaskedImage(p, mask=pixels[-1].astype(bool), copy=False)
else:
return Image(pixels, copy=False)
# Assumed not to have an Alpha channel
if normalize:
return Image(normalize_pixels_range(pixels), copy=False)
else:
return Image(pixels, copy=False)
def imageio_to_menpo(imio_reader, index):
pixels = imio_reader.get_data(index)
pixels = channels_to_front(pixels)
if pixels.shape[0] == 4:
# If normalise is False, then we return the alpha as an extra
# channel, which can be useful if the alpha channel has semantic
# meanings!
if normalise:
p = normalize_pixels_range(pixels[:3])
return MaskedImage(p, mask=pixels[-1].astype(np.bool),
copy=False)
else:
return Image(pixels, copy=False)
# Assumed not to have an Alpha channel
if normalise:
return Image(normalize_pixels_range(pixels), copy=False)
else:
return Image(pixels, copy=False)
def _read_one_frame(self):
r"""
Reads one frame from the opened ``self._pipe`` and converts it to
a numpy array
Returns
-------
image : :map:`Image`
Image of shape ``(self.height, self.width, 3)``
"""
raw_data = self._pipe.stdout.read(self.height * self.width * 3)
frame = np.frombuffer(raw_data, dtype=np.uint8)
frame = frame.reshape((self.height, self.width, 3))
self._pipe.stdout.flush()
self.index += 1
if self.normalize:
frame = normalize_pixels_range(frame)
return frame
def _read_one_frame(self):
r"""
Reads one frame from the opened ``self._pipe`` and converts it to
a numpy array
Returns
-------
image : :map:`Image`
Image of shape ``(self.height, self.width, 3)``
"""
raw_data = self._pipe.stdout.read(self.height*self.width*3)
frame = np.fromstring(raw_data, dtype=np.uint8)
frame = frame.reshape((self.height, self.width, 3))
self._pipe.stdout.flush()
self.index += 1
if self.normalize:
frame = normalize_pixels_range(frame)
return frame
def import_mesh(path, hasTexture=False, landmark_type='ibug68'):
if path.suffix == '.pkl' or path.suffix == '.gz':
mesh = import_pickle(path)
else:
mesh = m3io.import_mesh(path)
if hasTexture:
if mesh.texture.pixels.dtype != np.float64:
mesh.texture.pixels = normalize_pixels_range(mesh.texture.pixels)
else:
landmark = []
count = 0
with open(str(path) + '.landmark') as pp_file:
pp_file = csv.reader(pp_file, delimiter=' ')
for row in pp_file:
count = count + 1
if landmark_type == 'ibug100':
if count >= 1 and count <= 100:
landmark.append(
[float(row[0]),
float(row[1]),
float(row[2])])
if landmark_type == 'ibug68':
if count < 69:
landmark.append(
[float(row[0]),
float(row[1]),
float(row[2])])
if landmark_type == 'ibugEar':
if count >= 78 and count <= 88:
landmark.append(
[float(row[0]),
float(row[1]),
float(row[2])])
if count >= 90 and count <= 100:
landmark.append(
[float(row[0]),
float(row[1]),
float(row[2])])
mesh.landmarks[landmark_type] = PointCloud(np.array(landmark))
return mesh