#!/usr/bin/env python
# ******************************************************************************
# Copyright 2021 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.
# ******************************************************************************
"""
VGG model definition for ImageNet classification.
"""
from keras import Model, regularizers
from keras.layers import (Input, Activation, Dropout, Flatten, Rescaling)
from cnn2snn import quantize, load_quantized_model
from ..layer_blocks import conv_block, dense_block
from ..utils import fetch_file
BASE_WEIGHT_PATH = 'http://data.brainchip.com/models/vgg/'
[docs]def vgg_imagenet(input_shape=(224, 224, 3),
classes=1000,
include_top=True,
pooling=None,
weight_quantization=0,
activ_quantization=0,
input_weight_quantization=None,
input_scaling=(128, -1)):
"""Instantiates a VGG11 architecture with reduced number of filters in
convolutional layers (i.e. a quarter of the filters of the original
implementation of https://arxiv.org/pdf/1409.1556.pdf).
Args:
input_shape (tuple, optional): input shape tuple. Defaults to (224, 224,
3).
classes (int, optional): optional number of classes to classify images
into. Defaults to 1000.
include_top (bool, optional): whether to include the classification
layers 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.
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 VGG/ImageNet
"""
# check if overrides have been provided and override
if input_weight_quantization is None:
input_weight_quantization = weight_quantization
# Define weight regularization
weight_regularizer = regularizers.l2(4e-5)
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)
# Block 1
x = conv_block(x,
filters=16,
name='block_1/conv_1',
kernel_size=(3, 3),
padding='same',
kernel_regularizer=weight_regularizer,
add_batchnorm=True,
add_activation=True,
pooling='max',
pool_size=(2, 2))
# Block 2
x = conv_block(x,
filters=32,
name='block_2/conv_1',
kernel_size=(3, 3),
padding='same',
kernel_regularizer=weight_regularizer,
add_batchnorm=True,
add_activation=True,
pooling='max',
pool_size=(2, 2))
# Block 3
x = conv_block(x,
filters=64,
name='block_3/conv_1',
kernel_size=(3, 3),
padding='same',
kernel_regularizer=weight_regularizer,
add_batchnorm=True,
add_activation=True)
x = conv_block(x,
filters=64,
name='block_3/conv_2',
kernel_size=(3, 3),
padding='same',
kernel_regularizer=weight_regularizer,
add_batchnorm=True,
add_activation=True,
pooling='max',
pool_size=(2, 2))
# Block 4
x = conv_block(x,
filters=128,
name='block_4/conv_1',
kernel_size=(3, 3),
padding='same',
kernel_regularizer=weight_regularizer,
add_batchnorm=True,
add_activation=True)
x = conv_block(x,
filters=128,
name='block_4/conv_2',
kernel_size=(3, 3),
padding='same',
kernel_regularizer=weight_regularizer,
add_batchnorm=True,
add_activation=True,
pooling='max',
pool_size=(2, 2))
# Block 5
x = conv_block(x,
filters=128,
name='block_5/conv_1',
kernel_size=(3, 3),
padding='same',
kernel_regularizer=weight_regularizer,
add_batchnorm=True,
add_activation=True)
layer_pooling = 'global_avg' if pooling == 'avg' else 'max'
x = conv_block(x,
filters=128,
name='block_5/conv_2',
kernel_size=(3, 3),
padding='same',
kernel_regularizer=weight_regularizer,
add_batchnorm=True,
add_activation=True,
pooling=layer_pooling,
pool_size=(2, 2))
if include_top:
# Classification block
x = Flatten(name='flatten')(x)
x = dense_block(x,
units=4096,
name='fc_1',
add_batchnorm=True,
add_activation=True)
x = Dropout(0.5, name='dropout_1')(x)
x = dense_block(x,
units=4096,
name='fc_2',
add_batchnorm=True,
add_activation=True)
x = Dropout(0.5, name='dropout_2')(x)
x = dense_block(x,
units=classes,
name='predictions',
add_batchnorm=False,
add_activation=False)
act_function = 'softmax' if classes > 1 else 'sigmoid'
x = Activation(act_function, name=f'act_{act_function}')(x)
# Create model
model = Model(img_input, x, name='vgg11_%s_%s' % (input_shape[0], 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 vgg_imagenet_pretrained():
"""
Helper method to retrieve a `vgg_imagenet` model that was trained on
ImageNet dataset.
Returns:
keras.Model: a Keras Model instance.
"""
model_name = 'vgg11_imagenet_224_iq8_wq4_aq4.h5'
file_hash = '40fe5e7fdf083604d3bbe8eba314a832d1dc1f218a743bef244ff6ce2f3c9bfe'
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)