Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Refactoring of the cropping layers. #10865

Merged
merged 5 commits into from
Aug 8, 2018
Merged

Refactoring of the cropping layers. #10865

Changes from all commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
66095f3
Refactoring of the cropping layers.
gabrieldemarmiesse Aug 6, 2018
970169a
Refactoring of the cropping layers.
gabrieldemarmiesse Aug 6, 2018
f5c5521
Merge branch 'refactor_cropping_3' of github.com:gabrieldemarmiesse/k…
gabrieldemarmiesse Aug 7, 2018
a98a9ec
Merge branch 'master' into refactor_cropping_3
Aug 8, 2018
89acd6d
Changed the logic of the function to make it easier to read.
Aug 8, 2018
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
74 changes: 21 additions & 53 deletions keras/layers/convolutional.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2335,13 +2335,9 @@ def __init__(self, cropping=(1, 1), **kwargs):
self.input_spec = InputSpec(ndim=3)

def compute_output_shape(self, input_shape):
if input_shape[1] is not None:
length = input_shape[1] - self.cropping[0] - self.cropping[1]
else:
length = None
return (input_shape[0],
length,
input_shape[2])
return _compute_output_shape_cropping(input_shape,
'channels_last',
(self.cropping,))

def call(self, inputs):
return _call_cropping(inputs, 'channels_last', (self.cropping,))
Expand Down Expand Up @@ -2435,16 +2431,9 @@ def __init__(self, cropping=((0, 0), (0, 0)),
self.input_spec = InputSpec(ndim=4)

def compute_output_shape(self, input_shape):
if self.data_format == 'channels_first':
return (input_shape[0],
input_shape[1],
input_shape[2] - self.cropping[0][0] - self.cropping[0][1] if input_shape[2] else None,
input_shape[3] - self.cropping[1][0] - self.cropping[1][1] if input_shape[3] else None)
elif self.data_format == 'channels_last':
return (input_shape[0],
input_shape[1] - self.cropping[0][0] - self.cropping[0][1] if input_shape[1] else None,
input_shape[2] - self.cropping[1][0] - self.cropping[1][1] if input_shape[2] else None,
input_shape[3])
return _compute_output_shape_cropping(input_shape,
self.data_format,
self.cropping)

def call(self, inputs):
return _call_cropping(inputs, self.data_format, self.cropping)
Expand Down Expand Up @@ -2528,42 +2517,9 @@ def __init__(self, cropping=((1, 1), (1, 1), (1, 1)),
self.input_spec = InputSpec(ndim=5)

def compute_output_shape(self, input_shape):
if self.data_format == 'channels_first':
if input_shape[2] is not None:
dim1 = input_shape[2] - self.cropping[0][0] - self.cropping[0][1]
else:
dim1 = None
if input_shape[3] is not None:
dim2 = input_shape[3] - self.cropping[1][0] - self.cropping[1][1]
else:
dim2 = None
if input_shape[4] is not None:
dim3 = input_shape[4] - self.cropping[2][0] - self.cropping[2][1]
else:
dim3 = None
return (input_shape[0],
input_shape[1],
dim1,
dim2,
dim3)
elif self.data_format == 'channels_last':
if input_shape[1] is not None:
dim1 = input_shape[1] - self.cropping[0][0] - self.cropping[0][1]
else:
dim1 = None
if input_shape[2] is not None:
dim2 = input_shape[2] - self.cropping[1][0] - self.cropping[1][1]
else:
dim2 = None
if input_shape[3] is not None:
dim3 = input_shape[3] - self.cropping[2][0] - self.cropping[2][1]
else:
dim3 = None
return (input_shape[0],
dim1,
dim2,
dim3,
input_shape[4])
return _compute_output_shape_cropping(input_shape,
self.data_format,
self.cropping)

def call(self, inputs):
return _call_cropping(inputs, self.data_format, self.cropping)
Expand Down Expand Up @@ -2591,6 +2547,18 @@ def _call_cropping(inputs, data_format, cropping):
return inputs[slices]


def _compute_output_shape_cropping(input_shape, data_format, cropping):
cropping_all_dims = ((0, 0),) + cropping + ((0, 0),)
spatial_axes = list(range(1, 1 + len(cropping)))
cropping_all_dims = transpose_shape(cropping_all_dims, data_format, spatial_axes)

output_shape = list(input_shape)
for dim in range(len(output_shape)):
if output_shape[dim] is not None:
output_shape[dim] -= sum(cropping_all_dims[dim])
return tuple(output_shape)


# Aliases

Convolution1D = Conv1D
Expand Down