experiments/STT-tensorflow
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Export EfficientNet models to the public API.

Browse Source 
PiperOrigin-RevId: 306281184
Change-Id: I37257d5afdc0ffec6eacd218e8f2e86ad118fb89
This commit is contained in:
Francois Chollet 2020-04-13 12:06:13 -07:00 committed by TensorFlower Gardener
parent cb81df5708
commit 81594ed391
8 changed files with 237 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
tensorflow/python/keras/api/BUILD
View File

@ -17,6 +17,7 @@ keras_packages = [
"tensorflow.python.keras",
"tensorflow.python.keras.activations",
"tensorflow.python.keras.applications.densenet",
"tensorflow.python.keras.applications.efficientnet",
"tensorflow.python.keras.applications.imagenet_utils",
"tensorflow.python.keras.applications.inception_resnet_v2",
"tensorflow.python.keras.applications.inception_v3",

78
tensorflow/python/keras/applications/efficientnet.py
View File

@ -13,6 +13,7 @@
# limitations under the License.
# ==============================================================================
# pylint: disable=invalid-name
# pylint: disable=missing-docstring
"""EfficientNet models for Keras.
Reference paper:
@ -142,6 +143,53 @@ DENSE_KERNEL_INITIALIZER = {
layers = VersionAwareLayers()
BASE_DOCSTRING = """Instantiates the {name} architecture.
Reference paper:
- [EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks](
https://arxiv.org/abs/1905.11946) (ICML 2019)
Optionally loads weights pre-trained on ImageNet.
Note that the data format convention used by the model is
the one specified in your Keras config at `~/.keras/keras.json`.
If you have never configured it, it defaults to `"channels_last"`.
Arguments:
include_top: Whether to include the fully-connected
layer at the top of the network. Defaults to True.
weights: One of `None` (random initialization),
'imagenet' (pre-training on ImageNet),
or the path to the weights file to be loaded. Defaults to 'imagenet'.
input_tensor: Optional Keras tensor
(i.e. output of `layers.Input()`)
to use as image input for the model.
input_shape: Optional shape tuple, only to be specified
if `include_top` is False.
It should have exactly 3 inputs channels.
pooling: Optional pooling mode for feature extraction
when `include_top` is `False`. Defaults to None.
- `None` means that the output of the model will be
the 4D tensor output of the
last convolutional layer.
- `avg` means that global average pooling
will be applied to the output of the
last convolutional layer, and thus
the output of the model will be a 2D tensor.
- `max` means that global max pooling will
be applied.
classes: Optional number of classes to classify images
into, only to be specified if `include_top` is True, and
if no `weights` argument is specified. Defaults to 1000 (number of
ImageNet classes).
classifier_activation: A `str` or callable. The activation function to use
on the "top" layer. Ignored unless `include_top=True`. Set
`classifier_activation=None` to return the logits of the "top" layer.
Defaults to 'softmax'.
Returns:
A `keras.Model` instance.
"""
def EfficientNet(
width_coefficient,
@ -163,8 +211,8 @@ def EfficientNet(
"""Instantiates the EfficientNet architecture using given scaling coefficients.
Reference paper:
- [EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks]
(https://arxiv.org/abs/1905.11946) (ICML 2019)
- [EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks](
https://arxiv.org/abs/1905.11946) (ICML 2019)
Optionally loads weights pre-trained on ImageNet.
Note that the data format convention used by the model is
@ -474,6 +522,7 @@ def EfficientNetB0(include_top=True,
input_shape=None,
pooling=None,
classes=1000,
classifier_activation='softmax',
**kwargs):
return EfficientNet(
1.0,
@ -487,6 +536,7 @@ def EfficientNetB0(include_top=True,
input_shape=input_shape,
pooling=pooling,
classes=classes,
classifier_activation=classifier_activation,
**kwargs)
@ -498,6 +548,7 @@ def EfficientNetB1(include_top=True,
input_shape=None,
pooling=None,
classes=1000,
classifier_activation='softmax',
**kwargs):
return EfficientNet(
1.0,
@ -511,6 +562,7 @@ def EfficientNetB1(include_top=True,
input_shape=input_shape,
pooling=pooling,
classes=classes,
classifier_activation=classifier_activation,
**kwargs)
@ -522,6 +574,7 @@ def EfficientNetB2(include_top=True,
input_shape=None,
pooling=None,
classes=1000,
classifier_activation='softmax',
**kwargs):
return EfficientNet(
1.1,
@ -535,6 +588,7 @@ def EfficientNetB2(include_top=True,
input_shape=input_shape,
pooling=pooling,
classes=classes,
classifier_activation=classifier_activation,
**kwargs)
@ -546,6 +600,7 @@ def EfficientNetB3(include_top=True,
input_shape=None,
pooling=None,
classes=1000,
classifier_activation='softmax',
**kwargs):
return EfficientNet(
1.2,
@ -559,6 +614,7 @@ def EfficientNetB3(include_top=True,
input_shape=input_shape,
pooling=pooling,
classes=classes,
classifier_activation=classifier_activation,
**kwargs)
@ -570,6 +626,7 @@ def EfficientNetB4(include_top=True,
input_shape=None,
pooling=None,
classes=1000,
classifier_activation='softmax',
**kwargs):
return EfficientNet(
1.4,
@ -583,6 +640,7 @@ def EfficientNetB4(include_top=True,
input_shape=input_shape,
pooling=pooling,
classes=classes,
classifier_activation=classifier_activation,
**kwargs)
@ -594,6 +652,7 @@ def EfficientNetB5(include_top=True,
input_shape=None,
pooling=None,
classes=1000,
classifier_activation='softmax',
**kwargs):
return EfficientNet(
1.6,
@ -607,6 +666,7 @@ def EfficientNetB5(include_top=True,
input_shape=input_shape,
pooling=pooling,
classes=classes,
classifier_activation=classifier_activation,
**kwargs)
@ -618,6 +678,7 @@ def EfficientNetB6(include_top=True,
input_shape=None,
pooling=None,
classes=1000,
classifier_activation='softmax',
**kwargs):
return EfficientNet(
1.8,
@ -631,6 +692,7 @@ def EfficientNetB6(include_top=True,
input_shape=input_shape,
pooling=pooling,
classes=classes,
classifier_activation=classifier_activation,
**kwargs)
@ -642,6 +704,7 @@ def EfficientNetB7(include_top=True,
input_shape=None,
pooling=None,
classes=1000,
classifier_activation='softmax',
**kwargs):
return EfficientNet(
2.0,
@ -655,9 +718,20 @@ def EfficientNetB7(include_top=True,
input_shape=input_shape,
pooling=pooling,
classes=classes,
classifier_activation=classifier_activation,
**kwargs)
EfficientNetB0.__doc__ = BASE_DOCSTRING.format(name='EfficientNetB0')
EfficientNetB1.__doc__ = BASE_DOCSTRING.format(name='EfficientNetB1')
EfficientNetB2.__doc__ = BASE_DOCSTRING.format(name='EfficientNetB2')
EfficientNetB3.__doc__ = BASE_DOCSTRING.format(name='EfficientNetB3')
EfficientNetB4.__doc__ = BASE_DOCSTRING.format(name='EfficientNetB4')
EfficientNetB5.__doc__ = BASE_DOCSTRING.format(name='EfficientNetB5')
EfficientNetB6.__doc__ = BASE_DOCSTRING.format(name='EfficientNetB6')
EfficientNetB7.__doc__ = BASE_DOCSTRING.format(name='EfficientNetB7')
@keras_export('keras.applications.efficientnet.preprocess_input')
def preprocess_input(x, data_format=None): # pylint: disable=unused-argument
return x

1
tensorflow/python/tools/api/generator/api_init_files.bzl
View File

@ -83,6 +83,7 @@ KERAS_API_INIT_FILES = [
"keras/activations/__init__.py",
"keras/applications/__init__.py",
"keras/applications/densenet/__init__.py",
"keras/applications/efficientnet/__init__.py",
"keras/applications/imagenet_utils/__init__.py",
"keras/applications/inception_resnet_v2/__init__.py",
"keras/applications/inception_v3/__init__.py",

1
tensorflow/python/tools/api/generator/api_init_files_v1.bzl
View File

@ -103,6 +103,7 @@ KERAS_API_INIT_FILES_V1 = [
"keras/activations/__init__.py",
"keras/applications/__init__.py",
"keras/applications/densenet/__init__.py",
"keras/applications/efficientnet/__init__.py",
"keras/applications/imagenet_utils/__init__.py",
"keras/applications/inception_resnet_v2/__init__.py",
"keras/applications/inception_v3/__init__.py",

43
tensorflow/tools/api/golden/v1/tensorflow.keras.applications.efficientnet.pbtxt Normal file
View File

@ -0,0 +1,43 @@
path: "tensorflow.keras.applications.efficientnet"
tf_module {
member_method {
name: "EfficientNetB0"
argspec: "args=[\'include_top\', \'weights\', \'input_tensor\', \'input_shape\', \'pooling\', \'classes\', \'classifier_activation\'], varargs=None, keywords=kwargs, defaults=[\'True\', \'imagenet\', \'None\', \'None\', \'None\', \'1000\', \'softmax\'], "
}
member_method {
name: "EfficientNetB1"
argspec: "args=[\'include_top\', \'weights\', \'input_tensor\', \'input_shape\', \'pooling\', \'classes\', \'classifier_activation\'], varargs=None, keywords=kwargs, defaults=[\'True\', \'imagenet\', \'None\', \'None\', \'None\', \'1000\', \'softmax\'], "
}
member_method {
name: "EfficientNetB2"
argspec: "args=[\'include_top\', \'weights\', \'input_tensor\', \'input_shape\', \'pooling\', \'classes\', \'classifier_activation\'], varargs=None, keywords=kwargs, defaults=[\'True\', \'imagenet\', \'None\', \'None\', \'None\', \'1000\', \'softmax\'], "
}
member_method {
name: "EfficientNetB3"
argspec: "args=[\'include_top\', \'weights\', \'input_tensor\', \'input_shape\', \'pooling\', \'classes\', \'classifier_activation\'], varargs=None, keywords=kwargs, defaults=[\'True\', \'imagenet\', \'None\', \'None\', \'None\', \'1000\', \'softmax\'], "
}
member_method {
name: "EfficientNetB4"
argspec: "args=[\'include_top\', \'weights\', \'input_tensor\', \'input_shape\', \'pooling\', \'classes\', \'classifier_activation\'], varargs=None, keywords=kwargs, defaults=[\'True\', \'imagenet\', \'None\', \'None\', \'None\', \'1000\', \'softmax\'], "
}
member_method {
name: "EfficientNetB5"
argspec: "args=[\'include_top\', \'weights\', \'input_tensor\', \'input_shape\', \'pooling\', \'classes\', \'classifier_activation\'], varargs=None, keywords=kwargs, defaults=[\'True\', \'imagenet\', \'None\', \'None\', \'None\', \'1000\', \'softmax\'], "
}
member_method {
name: "EfficientNetB6"
argspec: "args=[\'include_top\', \'weights\', \'input_tensor\', \'input_shape\', \'pooling\', \'classes\', \'classifier_activation\'], varargs=None, keywords=kwargs, defaults=[\'True\', \'imagenet\', \'None\', \'None\', \'None\', \'1000\', \'softmax\'], "
}
member_method {
name: "EfficientNetB7"
argspec: "args=[\'include_top\', \'weights\', \'input_tensor\', \'input_shape\', \'pooling\', \'classes\', \'classifier_activation\'], varargs=None, keywords=kwargs, defaults=[\'True\', \'imagenet\', \'None\', \'None\', \'None\', \'1000\', \'softmax\'], "
}
member_method {
name: "decode_predictions"
argspec: "args=[\'preds\', \'top\'], varargs=None, keywords=None, defaults=[\'5\'], "
}
member_method {
name: "preprocess_input"
argspec: "args=[\'x\', \'data_format\'], varargs=None, keywords=None, defaults=[\'None\'], "
}
}

36
tensorflow/tools/api/golden/v1/tensorflow.keras.applications.pbtxt
View File

@ -4,6 +4,10 @@ tf_module {
name: "densenet"
mtype: "<type \'module\'>"
}
member {
name: "efficientnet"
mtype: "<type \'module\'>"
}
member {
name: "imagenet_utils"
mtype: "<type \'module\'>"
@ -64,6 +68,38 @@ tf_module {
name: "DenseNet201"
argspec: "args=[\'include_top\', \'weights\', \'input_tensor\', \'input_shape\', \'pooling\', \'classes\'], varargs=None, keywords=None, defaults=[\'True\', \'imagenet\', \'None\', \'None\', \'None\', \'1000\'], "
}
member_method {
name: "EfficientNetB0"
argspec: "args=[\'include_top\', \'weights\', \'input_tensor\', \'input_shape\', \'pooling\', \'classes\', \'classifier_activation\'], varargs=None, keywords=kwargs, defaults=[\'True\', \'imagenet\', \'None\', \'None\', \'None\', \'1000\', \'softmax\'], "
}
member_method {
name: "EfficientNetB1"
argspec: "args=[\'include_top\', \'weights\', \'input_tensor\', \'input_shape\', \'pooling\', \'classes\', \'classifier_activation\'], varargs=None, keywords=kwargs, defaults=[\'True\', \'imagenet\', \'None\', \'None\', \'None\', \'1000\', \'softmax\'], "
}
member_method {
name: "EfficientNetB2"
argspec: "args=[\'include_top\', \'weights\', \'input_tensor\', \'input_shape\', \'pooling\', \'classes\', \'classifier_activation\'], varargs=None, keywords=kwargs, defaults=[\'True\', \'imagenet\', \'None\', \'None\', \'None\', \'1000\', \'softmax\'], "
}
member_method {
name: "EfficientNetB3"
argspec: "args=[\'include_top\', \'weights\', \'input_tensor\', \'input_shape\', \'pooling\', \'classes\', \'classifier_activation\'], varargs=None, keywords=kwargs, defaults=[\'True\', \'imagenet\', \'None\', \'None\', \'None\', \'1000\', \'softmax\'], "
}
member_method {
name: "EfficientNetB4"
argspec: "args=[\'include_top\', \'weights\', \'input_tensor\', \'input_shape\', \'pooling\', \'classes\', \'classifier_activation\'], varargs=None, keywords=kwargs, defaults=[\'True\', \'imagenet\', \'None\', \'None\', \'None\', \'1000\', \'softmax\'], "
}
member_method {
name: "EfficientNetB5"
argspec: "args=[\'include_top\', \'weights\', \'input_tensor\', \'input_shape\', \'pooling\', \'classes\', \'classifier_activation\'], varargs=None, keywords=kwargs, defaults=[\'True\', \'imagenet\', \'None\', \'None\', \'None\', \'1000\', \'softmax\'], "
}
member_method {
name: "EfficientNetB6"
argspec: "args=[\'include_top\', \'weights\', \'input_tensor\', \'input_shape\', \'pooling\', \'classes\', \'classifier_activation\'], varargs=None, keywords=kwargs, defaults=[\'True\', \'imagenet\', \'None\', \'None\', \'None\', \'1000\', \'softmax\'], "
}
member_method {
name: "EfficientNetB7"
argspec: "args=[\'include_top\', \'weights\', \'input_tensor\', \'input_shape\', \'pooling\', \'classes\', \'classifier_activation\'], varargs=None, keywords=kwargs, defaults=[\'True\', \'imagenet\', \'None\', \'None\', \'None\', \'1000\', \'softmax\'], "
}
member_method {
name: "InceptionResNetV2"
argspec: "args=[\'include_top\', \'weights\', \'input_tensor\', \'input_shape\', \'pooling\', \'classes\', \'classifier_activation\'], varargs=None, keywords=kwargs, defaults=[\'True\', \'imagenet\', \'None\', \'None\', \'None\', \'1000\', \'softmax\'], "

43
tensorflow/tools/api/golden/v2/tensorflow.keras.applications.efficientnet.pbtxt Normal file
View File

@ -0,0 +1,43 @@
path: "tensorflow.keras.applications.efficientnet"
tf_module {
member_method {
name: "EfficientNetB0"
argspec: "args=[\'include_top\', \'weights\', \'input_tensor\', \'input_shape\', \'pooling\', \'classes\', \'classifier_activation\'], varargs=None, keywords=kwargs, defaults=[\'True\', \'imagenet\', \'None\', \'None\', \'None\', \'1000\', \'softmax\'], "
}
member_method {
name: "EfficientNetB1"
argspec: "args=[\'include_top\', \'weights\', \'input_tensor\', \'input_shape\', \'pooling\', \'classes\', \'classifier_activation\'], varargs=None, keywords=kwargs, defaults=[\'True\', \'imagenet\', \'None\', \'None\', \'None\', \'1000\', \'softmax\'], "
}
member_method {
name: "EfficientNetB2"
argspec: "args=[\'include_top\', \'weights\', \'input_tensor\', \'input_shape\', \'pooling\', \'classes\', \'classifier_activation\'], varargs=None, keywords=kwargs, defaults=[\'True\', \'imagenet\', \'None\', \'None\', \'None\', \'1000\', \'softmax\'], "
}
member_method {
name: "EfficientNetB3"
argspec: "args=[\'include_top\', \'weights\', \'input_tensor\', \'input_shape\', \'pooling\', \'classes\', \'classifier_activation\'], varargs=None, keywords=kwargs, defaults=[\'True\', \'imagenet\', \'None\', \'None\', \'None\', \'1000\', \'softmax\'], "
}
member_method {
name: "EfficientNetB4"
argspec: "args=[\'include_top\', \'weights\', \'input_tensor\', \'input_shape\', \'pooling\', \'classes\', \'classifier_activation\'], varargs=None, keywords=kwargs, defaults=[\'True\', \'imagenet\', \'None\', \'None\', \'None\', \'1000\', \'softmax\'], "
}
member_method {
name: "EfficientNetB5"
argspec: "args=[\'include_top\', \'weights\', \'input_tensor\', \'input_shape\', \'pooling\', \'classes\', \'classifier_activation\'], varargs=None, keywords=kwargs, defaults=[\'True\', \'imagenet\', \'None\', \'None\', \'None\', \'1000\', \'softmax\'], "
}
member_method {
name: "EfficientNetB6"
argspec: "args=[\'include_top\', \'weights\', \'input_tensor\', \'input_shape\', \'pooling\', \'classes\', \'classifier_activation\'], varargs=None, keywords=kwargs, defaults=[\'True\', \'imagenet\', \'None\', \'None\', \'None\', \'1000\', \'softmax\'], "
}
member_method {
name: "EfficientNetB7"
argspec: "args=[\'include_top\', \'weights\', \'input_tensor\', \'input_shape\', \'pooling\', \'classes\', \'classifier_activation\'], varargs=None, keywords=kwargs, defaults=[\'True\', \'imagenet\', \'None\', \'None\', \'None\', \'1000\', \'softmax\'], "
}
member_method {
name: "decode_predictions"
argspec: "args=[\'preds\', \'top\'], varargs=None, keywords=None, defaults=[\'5\'], "
}
member_method {
name: "preprocess_input"
argspec: "args=[\'x\', \'data_format\'], varargs=None, keywords=None, defaults=[\'None\'], "
}
}

36
tensorflow/tools/api/golden/v2/tensorflow.keras.applications.pbtxt
View File

@ -4,6 +4,10 @@ tf_module {
name: "densenet"
mtype: "<type \'module\'>"
}
member {
name: "efficientnet"
mtype: "<type \'module\'>"
}
member {
name: "imagenet_utils"
mtype: "<type \'module\'>"
@ -64,6 +68,38 @@ tf_module {
name: "DenseNet201"
argspec: "args=[\'include_top\', \'weights\', \'input_tensor\', \'input_shape\', \'pooling\', \'classes\'], varargs=None, keywords=None, defaults=[\'True\', \'imagenet\', \'None\', \'None\', \'None\', \'1000\'], "
}
member_method {
name: "EfficientNetB0"
argspec: "args=[\'include_top\', \'weights\', \'input_tensor\', \'input_shape\', \'pooling\', \'classes\', \'classifier_activation\'], varargs=None, keywords=kwargs, defaults=[\'True\', \'imagenet\', \'None\', \'None\', \'None\', \'1000\', \'softmax\'], "
}
member_method {
name: "EfficientNetB1"
argspec: "args=[\'include_top\', \'weights\', \'input_tensor\', \'input_shape\', \'pooling\', \'classes\', \'classifier_activation\'], varargs=None, keywords=kwargs, defaults=[\'True\', \'imagenet\', \'None\', \'None\', \'None\', \'1000\', \'softmax\'], "
}
member_method {
name: "EfficientNetB2"
argspec: "args=[\'include_top\', \'weights\', \'input_tensor\', \'input_shape\', \'pooling\', \'classes\', \'classifier_activation\'], varargs=None, keywords=kwargs, defaults=[\'True\', \'imagenet\', \'None\', \'None\', \'None\', \'1000\', \'softmax\'], "
}
member_method {
name: "EfficientNetB3"
argspec: "args=[\'include_top\', \'weights\', \'input_tensor\', \'input_shape\', \'pooling\', \'classes\', \'classifier_activation\'], varargs=None, keywords=kwargs, defaults=[\'True\', \'imagenet\', \'None\', \'None\', \'None\', \'1000\', \'softmax\'], "
}
member_method {
name: "EfficientNetB4"
argspec: "args=[\'include_top\', \'weights\', \'input_tensor\', \'input_shape\', \'pooling\', \'classes\', \'classifier_activation\'], varargs=None, keywords=kwargs, defaults=[\'True\', \'imagenet\', \'None\', \'None\', \'None\', \'1000\', \'softmax\'], "
}
member_method {
name: "EfficientNetB5"
argspec: "args=[\'include_top\', \'weights\', \'input_tensor\', \'input_shape\', \'pooling\', \'classes\', \'classifier_activation\'], varargs=None, keywords=kwargs, defaults=[\'True\', \'imagenet\', \'None\', \'None\', \'None\', \'1000\', \'softmax\'], "
}
member_method {
name: "EfficientNetB6"
argspec: "args=[\'include_top\', \'weights\', \'input_tensor\', \'input_shape\', \'pooling\', \'classes\', \'classifier_activation\'], varargs=None, keywords=kwargs, defaults=[\'True\', \'imagenet\', \'None\', \'None\', \'None\', \'1000\', \'softmax\'], "
}
member_method {
name: "EfficientNetB7"
argspec: "args=[\'include_top\', \'weights\', \'input_tensor\', \'input_shape\', \'pooling\', \'classes\', \'classifier_activation\'], varargs=None, keywords=kwargs, defaults=[\'True\', \'imagenet\', \'None\', \'None\', \'None\', \'1000\', \'softmax\'], "
}
member_method {
name: "InceptionResNetV2"
argspec: "args=[\'include_top\', \'weights\', \'input_tensor\', \'input_shape\', \'pooling\', \'classes\', \'classifier_activation\'], varargs=None, keywords=kwargs, defaults=[\'True\', \'imagenet\', \'None\', \'None\', \'None\', \'1000\', \'softmax\'], "