#!/usr/bin/env python
# ******************************************************************************
# Copyright 2020 Brainchip Holdings Ltd.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
# ******************************************************************************
"""
MobileNet model definition for ImageNet classification.
MobileNet V1 is a general architecture and can be used for multiple use cases.
This specific version includes parameter options to generate a mobilenet version
compatible for Akida with:
- overall architecture compatible with Akida (conv stride 2 replaced with
max pool),
- options to quantize weights and activations,
- different initialization options.
"""
from keras import Model, regularizers
from keras.layers import (Input, Reshape, Dropout, Activation, Rescaling)
from cnn2snn import load_quantized_model, quantize
from .imagenet_utils import obtain_input_shape
from ..layer_blocks import conv_block, separable_conv_block
from ..utils import fetch_file
BASE_WEIGHT_PATH = 'http://data.brainchip.com/models/mobilenet/'
[docs]def mobilenet_imagenet(input_shape=None,
alpha=1.0,
dropout=1e-3,
include_top=True,
pooling=None,
classes=1000,
use_stride2=False,
weight_quantization=0,
activ_quantization=0,
input_weight_quantization=None,
input_scaling=(128, -1)):
"""Instantiates the MobileNet architecture.
Args:
input_shape (tuple, optional): shape tuple. Defaults to None.
alpha (float, optional): controls the width of the model.
Defaults to 1.0.
* If `alpha` < 1.0, proportionally decreases the number of filters
in each layer.
* If `alpha` > 1.0, proportionally increases the number of filters
in each layer.
* If `alpha` = 1, default number of filters from the paper are used
at each layer.
dropout (float, optional): dropout rate. Defaults to 1e-3.
include_top (bool, optional): whether to include the fully-connected
layer at the top of the model. Defaults to True.
pooling (str, optional): 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 block.
* `avg` means that global average pooling will be applied to the
output of the last convolutional block, and thus the output of the
model will be a 2D tensor.
classes (int, optional): optional number of classes to classify images
into, only to be specified if `include_top` is `True`. Defaults to 1000.
use_stride2 (bool, optional): replace max pooling operations by stride 2
convolutions in layers separable 2, 4, 6 and 12. Defaults to False.
weight_quantization (int, optional): sets all weights in the model to have
a particular quantization bitwidth except for the weights in the
first layer.
Defaults to 0.
* '0' implements floating point 32-bit weights.
* '2' through '8' implements n-bit weights where n is from 2-8 bits.
activ_quantization (int, optional): sets all activations in the model to have a
particular activation quantization bitwidth.
Defaults to 0.
* '0' implements floating point 32-bit activations.
* '2' through '8' implements n-bit weights where n is from 2-8 bits.
input_weight_quantization (int, optional): sets weight quantization in the first layer.
Defaults to weight_quantization value.
* '0' implements floating point 32-bit weights.
* '2' through '8' implements n-bit weights where n is from 2-8 bits.
input_scaling (tuple, optional): scale factor and offset to apply to
inputs. Defaults to (128, -1). Note that following Akida convention,
the scale factor is an integer used as a divider.
Returns:
keras.Model: a Keras model for MobileNet/ImageNet.
Raises:
ValueError: in case of invalid input shape.
"""
# check if overrides have been provided and override
if input_weight_quantization is None:
input_weight_quantization = weight_quantization
# Define weight regularization, will apply to the first convolutional layer
# and to all pointwise weights of separable convolutional layers.
weight_regularizer = regularizers.l2(4e-5)
# Define stride 2 or max pooling
if use_stride2:
sep_conv_pooling = None
strides = 2
else:
sep_conv_pooling = 'max'
strides = 1
# Determine proper input shape and default size.
if input_shape is None:
default_size = 224
else:
rows = input_shape[0]
cols = input_shape[1]
if rows == cols and rows in [128, 160, 192, 224]:
default_size = rows
else:
default_size = 224
input_shape = obtain_input_shape(input_shape,
default_size=default_size,
min_size=32,
include_top=include_top)
rows = input_shape[0]
cols = input_shape[1]
img_input = Input(shape=input_shape, name="input")
if input_scaling is None:
x = img_input
else:
scale, offset = input_scaling
x = Rescaling(1. / scale, offset, name="rescaling")(img_input)
x = conv_block(x,
filters=int(32 * alpha),
name='conv_0',
kernel_size=(3, 3),
padding='same',
use_bias=False,
strides=2,
add_batchnorm=True,
add_activation=True,
kernel_regularizer=weight_regularizer)
x = separable_conv_block(x,
filters=int(64 * alpha),
name='separable_1',
kernel_size=(3, 3),
padding='same',
use_bias=False,
add_batchnorm=True,
add_activation=True,
pointwise_regularizer=weight_regularizer)
x = separable_conv_block(x,
filters=int(128 * alpha),
name='separable_2',
kernel_size=(3, 3),
padding='same',
pooling=sep_conv_pooling,
strides=strides,
use_bias=False,
add_batchnorm=True,
add_activation=True,
pointwise_regularizer=weight_regularizer)
x = separable_conv_block(x,
filters=int(128 * alpha),
name='separable_3',
kernel_size=(3, 3),
padding='same',
use_bias=False,
add_batchnorm=True,
add_activation=True,
pointwise_regularizer=weight_regularizer)
x = separable_conv_block(x,
filters=int(256 * alpha),
name='separable_4',
kernel_size=(3, 3),
padding='same',
pooling=sep_conv_pooling,
strides=strides,
use_bias=False,
add_batchnorm=True,
add_activation=True,
pointwise_regularizer=weight_regularizer)
x = separable_conv_block(x,
filters=int(256 * alpha),
name='separable_5',
kernel_size=(3, 3),
padding='same',
use_bias=False,
add_batchnorm=True,
add_activation=True,
pointwise_regularizer=weight_regularizer)
x = separable_conv_block(x,
filters=int(512 * alpha),
name='separable_6',
kernel_size=(3, 3),
padding='same',
pooling=sep_conv_pooling,
strides=strides,
use_bias=False,
add_batchnorm=True,
add_activation=True,
pointwise_regularizer=weight_regularizer)
x = separable_conv_block(x,
filters=int(512 * alpha),
name='separable_7',
kernel_size=(3, 3),
padding='same',
use_bias=False,
add_batchnorm=True,
add_activation=True,
pointwise_regularizer=weight_regularizer)
x = separable_conv_block(x,
filters=int(512 * alpha),
name='separable_8',
kernel_size=(3, 3),
padding='same',
use_bias=False,
add_batchnorm=True,
add_activation=True,
pointwise_regularizer=weight_regularizer)
x = separable_conv_block(x,
filters=int(512 * alpha),
name='separable_9',
kernel_size=(3, 3),
padding='same',
use_bias=False,
add_batchnorm=True,
add_activation=True,
pointwise_regularizer=weight_regularizer)
x = separable_conv_block(x,
filters=int(512 * alpha),
name='separable_10',
kernel_size=(3, 3),
padding='same',
use_bias=False,
add_batchnorm=True,
add_activation=True,
pointwise_regularizer=weight_regularizer)
x = separable_conv_block(x,
filters=int(512 * alpha),
name='separable_11',
kernel_size=(3, 3),
padding='same',
use_bias=False,
add_batchnorm=True,
add_activation=True,
pointwise_regularizer=weight_regularizer)
x = separable_conv_block(x,
filters=int(1024 * alpha),
name='separable_12',
kernel_size=(3, 3),
padding='same',
pooling=sep_conv_pooling,
strides=strides,
use_bias=False,
add_batchnorm=True,
add_activation=True,
pointwise_regularizer=weight_regularizer)
# Last separable layer with global pooling
layer_pooling = 'global_avg' if include_top or pooling == 'avg' else None
x = separable_conv_block(x,
filters=int(1024 * alpha),
name='separable_13',
kernel_size=(3, 3),
padding='same',
pooling=layer_pooling,
use_bias=False,
add_batchnorm=True,
add_activation=True,
pointwise_regularizer=weight_regularizer)
if include_top:
shape = (1, 1, int(1024 * alpha))
x = Reshape(shape, name='reshape_1')(x)
x = Dropout(dropout, name='dropout')(x)
x = separable_conv_block(x,
filters=classes,
name='separable_14',
kernel_size=(3, 3),
padding='same',
use_bias=False,
add_batchnorm=False,
add_activation=False,
pointwise_regularizer=weight_regularizer)
act_function = 'softmax' if classes > 1 else 'sigmoid'
x = Activation(act_function, name=f'act_{act_function}')(x)
x = Reshape((classes,), name='reshape_2')(x)
# Create model.
model = Model(img_input,
x,
name='mobilenet_%0.2f_%s_%s' % (alpha, rows, classes))
if ((weight_quantization != 0) or (activ_quantization != 0) or
(input_weight_quantization != 0)):
return quantize(model, weight_quantization, activ_quantization,
input_weight_quantization)
return model
[docs]def mobilenet_imagenet_pretrained(alpha=1.0):
"""
Helper method to retrieve a `mobilenet_imagenet` model that was trained on
ImageNet dataset.
Args:
alpha (float): width of the model.
Returns:
keras.Model: a Keras Model instance.
"""
if alpha == 1.0:
model_name = 'mobilenet_imagenet_224_iq8_wq4_aq4.h5'
file_hash = '242a5d6e155f9677e4a295d6c0b092061ab442bd2c51898663d818b706fc2c34'
elif alpha == 0.5:
model_name = 'mobilenet_imagenet_224_alpha_50_iq8_wq4_aq4.h5'
file_hash = 'aa8fec3c21455dbb16dc86063ea07fcba792dcdaf8293c58db42dc379c78afc1'
elif alpha == 0.25:
model_name = 'mobilenet_imagenet_224_alpha_25_iq8_wq4_aq4.h5'
file_hash = '2765b8b24e880f593dbb18d73e8750e51a3dfc8f38f2978d9d6679e3f970466a'
else:
raise ValueError(
f"Requested model with alpha={alpha} is not available.")
model_path = fetch_file(BASE_WEIGHT_PATH + model_name,
fname=model_name,
file_hash=file_hash,
cache_subdir='models')
return load_quantized_model(model_path)