平時經(jīng)常使用作為基礎(chǔ)網(wǎng)絡(luò)，但是對它網(wǎng)絡(luò)結(jié)構(gòu)的了解卻一直沒有太深入質(zhì)量效應(yīng)3男主默認代碼，所以就想著通過閱讀代碼來達到對該網(wǎng)絡(luò)結(jié)構(gòu)加深了解的目的。
我們以 V3為例，看代碼之前先了解下 的結(jié)構(gòu)特點：
最開始的7層和最后2層是普通的網(wǎng)絡(luò)結(jié)構(gòu)，最大的區(qū)別在于中間的三個 模塊組，這個模塊組使用了多種尺寸卷積核的并聯(lián)結(jié)構(gòu)；大幅使用1*1的卷積核；多次使用小卷積核代替更大尺寸的卷積核；這種結(jié)構(gòu)在前幾層效果不太好，但對特征圖大小為12~20的中間層效果明顯；使用全局平均池化代替最后的全連接層。
谷歌的官方代碼如下，我會在代碼上方加上自己的理解注釋：
# Copyright 2016 The TensorFlow Authors. All Rights Reserved.
#
# 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.
# ==============================================================================
"""Contains the definition for inception v3 classification network."""
from __future__ import absolute_import
from __future__ import division
from __future__ import print_function
import tensorflow as tf
from nets import inception_utils
slim = tf.contrib.slim
trunc_normal = lambda stddev: tf.truncated_normal_initializer(0.0, stddev)
# 定義了 inception 網(wǎng)絡(luò)從輸入到輸出前面一層的網(wǎng)絡(luò)結(jié)構(gòu)，
# 在 inception_v3 的主體的開頭部分調(diào)用
def inception_v3_base(inputs,
final_endpoint='Mixed_7c',
min_depth=16,
depth_multiplier=1.0,
scope=None):
"""Inception model from http://arxiv.org/abs/1512.00567.
Constructs an Inception v3 network from inputs to the given final endpoint.
This method can construct the network up to the final inception block
Mixed_7c.
Note that the names of the layers in the paper do not correspond to the names
of the endpoints registered by this function although they build the same
network.
Here is a mapping from the old_names to the new names:
Old name | New name 網(wǎng)絡(luò)層的名稱映射
=======================================
從這里可以很清晰的看到網(wǎng)絡(luò)的結(jié)構(gòu)，前面的7個普通網(wǎng)絡(luò)層，其中包括5個卷積層和2個池化層，
后接三個 mixed 模塊，最后是 inception_v3 主體中的輸出層
conv0 | Conv2d_1a_3x3
conv1 | Conv2d_2a_3x3
conv2 | Conv2d_2b_3x3
pool1 | MaxPool_3a_3x3
conv3 | Conv2d_3b_1x1
conv4 | Conv2d_4a_3x3
pool2 | MaxPool_5a_3x3
mixed_35x35x256a | Mixed_5b
mixed_35x35x288a | Mixed_5c
mixed_35x35x288b | Mixed_5d
mixed_17x17x768a | Mixed_6a
mixed_17x17x768b | Mixed_6b
mixed_17x17x768c | Mixed_6c
mixed_17x17x768d | Mixed_6d
mixed_17x17x768e | Mixed_6e
mixed_8x8x1280a | Mixed_7a
mixed_8x8x2048a | Mixed_7b
mixed_8x8x2048b | Mixed_7c
Args:
輸入的 tensor, 形狀為 [batch_size, height, width, channels]
inputs: a tensor of size [batch_size, height, width, channels].
這個 base 模塊的終點節(jié)點名稱，可以是該網(wǎng)絡(luò)的任意一層
final_endpoint: specifies the endpoint to construct the network up to. It
can be one of ['Conv2d_1a_3x3', 'Conv2d_2a_3x3', 'Conv2d_2b_3x3',
'MaxPool_3a_3x3', 'Conv2d_3b_1x1', 'Conv2d_4a_3x3', 'MaxPool_5a_3x3',
'Mixed_5b', 'Mixed_5c', 'Mixed_5d', 'Mixed_6a', 'Mixed_6b', 'Mixed_6c',
'Mixed_6d', 'Mixed_6e', 'Mixed_7a', 'Mixed_7b', 'Mixed_7c'].
卷積層的最小通道數(shù)， 這里設(shè)置為16
min_depth: Minimum depth value (number of channels) for all convolution ops.
Enforced when depth_multiplier < 1, and not an active constraint when
depth_multiplier >= 1.
通道數(shù)的倍數(shù)，設(shè)置范圍為0-1，可以設(shè)置小點，減少計算量，默認為1
depth_multiplier: Float multiplier for the depth (number of channels)
for all convolution ops. The value must be greater than zero. Typical
usage will be to set this value in (0, 1) to reduce the number of
parameters or computation cost of the model.
模型的變量空間
scope: Optional variable_scope.
Returns:
網(wǎng)絡(luò)輸出的張量
tensor_out: output tensor corresponding to the final_endpoint.
輔助輸出
end_points: a set of activations for external use, for example summaries or
losses.
Raises:
ValueError: if final_endpoint is not set to one of the predefined values,
or depth_multiplier <= 0
"""
# end_points will collect relevant activations for external use, for example
# summaries or losses.
end_points = {}
# 通道數(shù)倍數(shù)不能小于0，否則報錯
if depth_multiplier <= 0:
raise ValueError('depth_multiplier is not greater than zero.')
# 每個卷積層的通道數(shù)計算公式， 比如，倍數(shù)設(shè)置為2，輸入32，則輸出通道數(shù)為64，如果小于16，則輸出16
depth = lambda d: max(int(d * depth_multiplier), min_depth)
with tf.variable_scope(scope, 'InceptionV3', [inputs]):
with slim.arg_scope([slim.conv2d, slim.max_pool2d, slim.avg_pool2d],
stride=1, padding='VALID'): # padding='VALID'！！！
# 299 x 299 x 3 輸入圖片的尺寸，因為stride為2，所以輸出尺寸減半
end_point = 'Conv2d_1a_3x3'
net = slim.conv2d(inputs, depth(32), [3, 3], stride=2, scope=end_point)
end_points[end_point] = net # 增加輔助節(jié)點信息
# 如果網(wǎng)絡(luò)終點層在這里，就在這里停止，并且返回，后續(xù)都是如此
if end_point == final_endpoint: return net, end_points
# 149 x 149 x 32 默認的padding方式為 valid，不填充，導(dǎo)致尺寸減少了2
end_point = 'Conv2d_2a_3x3'
net = slim.conv2d(net, depth(32), [3, 3], scope=end_point)
end_points[end_point] = net
if end_point == final_endpoint: return net, end_points
# 147 x 147 x 32 padding='SAME' 尺寸不變
end_point = 'Conv2d_2b_3x3'
net = slim.conv2d(net, depth(64), [3, 3], padding='SAME', scope=end_point)
end_points[end_point] = net
if end_point == final_endpoint: return net, end_points
# 147 x 147 x 64 池化層， stride=2 ，尺寸減半
end_point = 'MaxPool_3a_3x3'
net = slim.max_pool2d(net, [3, 3], stride=2, scope=end_point)
end_points[end_point] = net
if end_point == final_endpoint: return net, end_points
# 73 x 73 x 64 卷積核為[1, 1]，尺寸不變
end_point = 'Conv2d_3b_1x1'
net = slim.conv2d(net, depth(80), [1, 1], scope=end_point)
end_points[end_point] = net
if end_point == final_endpoint: return net, end_points
# 73 x 73 x 80. 默認的padding方式為 valid，不填充，導(dǎo)致尺寸減少了2
end_point = 'Conv2d_4a_3x3'
net = slim.conv2d(net, depth(192), [3, 3], scope=end_point)
end_points[end_point] = net
if end_point == final_endpoint: return net, end_points
# 71 x 71 x 192. 池化層， stride=2 ，尺寸減半
end_point = 'MaxPool_5a_3x3'
net = slim.max_pool2d(net, [3, 3], stride=2, scope=end_point)
end_points[end_point] = net
if end_point == final_endpoint: return net, end_points
# 35 x 35 x 192. 前面7層最后的輸出尺寸
# Inception blocks inception 的并聯(lián)模塊
with slim.arg_scope([slim.conv2d, slim.max_pool2d, slim.avg_pool2d],
stride=1, padding='SAME'):
# 注意這里的參數(shù)都是 stride=1, padding='SAME' padding改變了
# mixed: 35 x 35 x 256.
end_point = 'Mixed_5b'
with tf.variable_scope(end_point):
with tf.variable_scope('Branch_0'):
# 這里得到35 x 35 x 64
branch_0 = slim.conv2d(net, depth(64), [1, 1], scope='Conv2d_0a_1x1')
with tf.variable_scope('Branch_1'):
# 這里得到 35 x 35 x 64
branch_1 = slim.conv2d(net, depth(48), [1, 1], scope='Conv2d_0a_1x1')
branch_1 = slim.conv2d(branch_1, depth(64), [5, 5],
scope='Conv2d_0b_5x5')
with tf.variable_scope('Branch_2'):
# 這里得到 35 x 35 x 96
branch_2 = slim.conv2d(net, depth(64), [1, 1], scope='Conv2d_0a_1x1')
branch_2 = slim.conv2d(branch_2, depth(96), [3, 3],
scope='Conv2d_0b_3x3')
branch_2 = slim.conv2d(branch_2, depth(96), [3, 3],
scope='Conv2d_0c_3x3')
with tf.variable_scope('Branch_3'):
# 這里得到 35 x 35 x 32
branch_3 = slim.avg_pool2d(net, [3, 3], scope='AvgPool_0a_3x3')
branch_3 = slim.conv2d(branch_3, depth(32), [1, 1],
scope='Conv2d_0b_1x1')
# 最后第三維度相加，64+64+96+32=256， 最后輸出的尺寸為 35 x 35 x 256
# 這里非常明顯的體現(xiàn)了 inception 的并聯(lián)思路， 1*1，1*1和5*5， 1*1和3*3和3*3，平均池化和1*1，
# 這里四個支路的并聯(lián)，最后拼接在一起
net = tf.concat(axis=3, values=[branch_0, branch_1, branch_2, branch_3])
end_points[end_point] = net
if end_point == final_endpoint: return net, end_points
# mixed_1: 35 x 35 x 288. 這里和上面的模塊就是一個思路了，簡單的做一個注釋
end_point = 'Mixed_5c'
with tf.variable_scope(end_point):
with tf.variable_scope('Branch_0'):
# 這里得到 35 x 35 x 64
branch_0 = slim.conv2d(net, depth(64), [1, 1], scope='Conv2d_0a_1x1')
with tf.variable_scope('Branch_1'):
# 這里得到 35 x 35 x 64
branch_1 = slim.conv2d(net, depth(48), [1, 1], scope='Conv2d_0b_1x1')
branch_1 = slim.conv2d(branch_1, depth(64), [5, 5],
scope='Conv_1_0c_5x5')
with tf.variable_scope('Branch_2'):
# 這里得到 35 x 35 x 96
branch_2 = slim.conv2d(net, depth(64), [1, 1],
scope='Conv2d_0a_1x1')
branch_2 = slim.conv2d(branch_2, depth(96), [3, 3],
scope='Conv2d_0b_3x3')
branch_2 = slim.conv2d(branch_2, depth(96), [3, 3],
scope='Conv2d_0c_3x3')
with tf.variable_scope('Branch_3'):
# 這里得到 35 x 35 x 64 比之前的32多了32
branch_3 = slim.avg_pool2d(net, [3, 3], scope='AvgPool_0a_3x3')
branch_3 = slim.conv2d(branch_3, depth(64), [1, 1],
scope='Conv2d_0b_1x1')
# 最后第三維度相加，64+64+96+64=288， 最后輸出的尺寸為 35 x 35 x 288
net = tf.concat(axis=3, values=[branch_0, branch_1, branch_2, branch_3])
end_points[end_point] = net
if end_point == final_endpoint: return net, end_points
# mixed_2: 35 x 35 x 288. 這個模塊和上面的模塊一模一樣
end_point = 'Mixed_5d'
with tf.variable_scope(end_point):
with tf.variable_scope('Branch_0'):
branch_0 = slim.conv2d(net, depth(64), [1, 1], scope='Conv2d_0a_1x1')
with tf.variable_scope('Branch_1'):
branch_1 = slim.conv2d(net, depth(48), [1, 1], scope='Conv2d_0a_1x1')
branch_1 = slim.conv2d(branch_1, depth(64), [5, 5],
scope='Conv2d_0b_5x5')
with tf.variable_scope('Branch_2'):
branch_2 = slim.conv2d(net, depth(64), [1, 1], scope='Conv2d_0a_1x1')
branch_2 = slim.conv2d(branch_2, depth(96), [3, 3],
scope='Conv2d_0b_3x3')
branch_2 = slim.conv2d(branch_2, depth(96), [3, 3],
scope='Conv2d_0c_3x3')
with tf.variable_scope('Branch_3'):
branch_3 = slim.avg_pool2d(net, [3, 3], scope='AvgPool_0a_3x3')
branch_3 = slim.conv2d(branch_3, depth(64), [1, 1],
scope='Conv2d_0b_1x1')
net = tf.concat(axis=3, values=[branch_0, branch_1, branch_2, branch_3])
end_points[end_point] = net
if end_point == final_endpoint: return net, end_points
# mixed_3: 17 x 17 x 768. 新的并聯(lián)結(jié)構(gòu)模塊，上面的最后輸出的尺寸為 35 x 35 x 288
end_point = 'Mixed_6a'
with tf.variable_scope(end_point):
with tf.variable_scope('Branch_0'):
# 這里得到 17 x 17 x 384 因為 stride=2
branch_0 = slim.conv2d(net, depth(384), [3, 3], stride=2,
padding='VALID', scope='Conv2d_1a_1x1')
with tf.variable_scope('Branch_1'):
# 這里得到 17 x 17 x 96 因為 stride=2
branch_1 = slim.conv2d(net, depth(64), [1, 1], scope='Conv2d_0a_1x1')
branch_1 = slim.conv2d(branch_1, depth(96), [3, 3],
scope='Conv2d_0b_3x3')
branch_1 = slim.conv2d(branch_1, depth(96), [3, 3], stride=2,
padding='VALID', scope='Conv2d_1a_1x1')
with tf.variable_scope('Branch_2'):
# 這里得到 17 x 17 x 288 因為 stride=2
branch_2 = slim.max_pool2d(net, [3, 3], stride=2, padding='VALID',
scope='MaxPool_1a_3x3')
# 384+96+288=768
# 之前的并聯(lián)思路， 1*1，1*1和5*5， 1*1和3*3和3*3，平均池化和1*1， 35 x 35 x 288
# 這里的并聯(lián)思路， 3*3， 1*1和3*3和3*3，最大池化，stride=2，17 x 17 x 768， 通道數(shù)增加到768
net = tf.concat(axis=3, values=[branch_0, branch_1, branch_2])
end_points[end_point] = net
if end_point == final_endpoint: return net, end_points
# mixed4: 17 x 17 x 768.
# 之前的并聯(lián)思路， 1*1，1*1和5*5， 1*1和3*3和3*3，平均池化和1*1， 35 x 35 x 288
# 之前的并聯(lián)思路， 3*3， 1*1和3*3和3*3，最大池化，stride=2，17 x 17 x 768， 通道數(shù)增加到768
# 這里的并聯(lián)思路， 1*1，1*1和1*7和7*1， 1*1和1*7和7*1和1*7和7*1，平均池化和1*1， 17 x 17 x 768， 通道數(shù)增加到768
end_point = 'Mixed_6b'
with tf.variable_scope(end_point):
with tf.variable_scope('Branch_0'):
branch_0 = slim.conv2d(net, depth(192), [1, 1], scope='Conv2d_0a_1x1')
with tf.variable_scope('Branch_1'):
branch_1 = slim.conv2d(net, depth(128), [1, 1], scope='Conv2d_0a_1x1')
branch_1 = slim.conv2d(branch_1, depth(128), [1, 7],
scope='Conv2d_0b_1x7')
branch_1 = slim.conv2d(branch_1, depth(192), [7, 1],
scope='Conv2d_0c_7x1')
with tf.variable_scope('Branch_2'):
branch_2 = slim.conv2d(net, depth(128), [1, 1], scope='Conv2d_0a_1x1')
branch_2 = slim.conv2d(branch_2, depth(128), [7, 1],
scope='Conv2d_0b_7x1')
branch_2 = slim.conv2d(branch_2, depth(128), [1, 7],
scope='Conv2d_0c_1x7')
branch_2 = slim.conv2d(branch_2, depth(128), [7, 1],
scope='Conv2d_0d_7x1')
branch_2 = slim.conv2d(branch_2, depth(192), [1, 7],
scope='Conv2d_0e_1x7')
with tf.variable_scope('Branch_3'):
branch_3 = slim.avg_pool2d(net, [3, 3], scope='AvgPool_0a_3x3')
branch_3 = slim.conv2d(branch_3, depth(192), [1, 1],
scope='Conv2d_0b_1x1')
net = tf.concat(axis=3, values=[branch_0, branch_1, branch_2, branch_3])
end_points[end_point] = net
if end_point == final_endpoint: return net, end_points
# mixed_5: 17 x 17 x 768. 這個和上面的模塊一模一樣
end_point = 'Mixed_6c'
with tf.variable_scope(end_point):
with tf.variable_scope('Branch_0'):
branch_0 = slim.conv2d(net, depth(192), [1, 1], scope='Conv2d_0a_1x1')
with tf.variable_scope('Branch_1'):
branch_1 = slim.conv2d(net, depth(160), [1, 1], scope='Conv2d_0a_1x1')
branch_1 = slim.conv2d(branch_1, depth(160), [1, 7],
scope='Conv2d_0b_1x7')
branch_1 = slim.conv2d(branch_1, depth(192), [7, 1],
scope='Conv2d_0c_7x1')
with tf.variable_scope('Branch_2'):
branch_2 = slim.conv2d(net, depth(160), [1, 1], scope='Conv2d_0a_1x1')
branch_2 = slim.conv2d(branch_2, depth(160), [7, 1],
scope='Conv2d_0b_7x1')
branch_2 = slim.conv2d(branch_2, depth(160), [1, 7],
scope='Conv2d_0c_1x7')
branch_2 = slim.conv2d(branch_2, depth(160), [7, 1],
scope='Conv2d_0d_7x1')
branch_2 = slim.conv2d(branch_2, depth(192), [1, 7],
scope='Conv2d_0e_1x7')
with tf.variable_scope('Branch_3'):
branch_3 = slim.avg_pool2d(net, [3, 3], scope='AvgPool_0a_3x3')
branch_3 = slim.conv2d(branch_3, depth(192), [1, 1],
scope='Conv2d_0b_1x1')
net = tf.concat(axis=3, values=[branch_0, branch_1, branch_2, branch_3])
end_points[end_point] = net
if end_point == final_endpoint: return net, end_points
# mixed_6: 17 x 17 x 768. 這個和上面的模塊一模一樣
end_point = 'Mixed_6d'
with tf.variable_scope(end_point):
with tf.variable_scope('Branch_0'):
branch_0 = slim.conv2d(net, depth(192), [1, 1], scope='Conv2d_0a_1x1')
with tf.variable_scope('Branch_1'):
branch_1 = slim.conv2d(net, depth(160), [1, 1], scope='Conv2d_0a_1x1')
branch_1 = slim.conv2d(branch_1, depth(160), [1, 7],
scope='Conv2d_0b_1x7')
branch_1 = slim.conv2d(branch_1, depth(192), [7, 1],
scope='Conv2d_0c_7x1')
with tf.variable_scope('Branch_2'):
branch_2 = slim.conv2d(net, depth(160), [1, 1], scope='Conv2d_0a_1x1')
branch_2 = slim.conv2d(branch_2, depth(160), [7, 1],
scope='Conv2d_0b_7x1')
branch_2 = slim.conv2d(branch_2, depth(160), [1, 7],
scope='Conv2d_0c_1x7')
branch_2 = slim.conv2d(branch_2, depth(160), [7, 1],
scope='Conv2d_0d_7x1')
branch_2 = slim.conv2d(branch_2, depth(192), [1, 7],
scope='Conv2d_0e_1x7')
with tf.variable_scope('Branch_3'):
branch_3 = slim.avg_pool2d(net, [3, 3], scope='AvgPool_0a_3x3')
branch_3 = slim.conv2d(branch_3, depth(192), [1, 1],
scope='Conv2d_0b_1x1')
net = tf.concat(axis=3, values=[branch_0, branch_1, branch_2, branch_3])
end_points[end_point] = net
if end_point == final_endpoint: return net, end_points
# mixed_7: 17 x 17 x 768. 這個和上面的模塊一模一樣
end_point = 'Mixed_6e'
with tf.variable_scope(end_point):
with tf.variable_scope('Branch_0'):
branch_0 = slim.conv2d(net, depth(192), [1, 1], scope='Conv2d_0a_1x1')
with tf.variable_scope('Branch_1'):
branch_1 = slim.conv2d(net, depth(192), [1, 1], scope='Conv2d_0a_1x1')
branch_1 = slim.conv2d(branch_1, depth(192), [1, 7],
scope='Conv2d_0b_1x7')
branch_1 = slim.conv2d(branch_1, depth(192), [7, 1],
scope='Conv2d_0c_7x1')
with tf.variable_scope('Branch_2'):
branch_2 = slim.conv2d(net, depth(192), [1, 1], scope='Conv2d_0a_1x1')
branch_2 = slim.conv2d(branch_2, depth(192), [7, 1],
scope='Conv2d_0b_7x1')
branch_2 = slim.conv2d(branch_2, depth(192), [1, 7],
scope='Conv2d_0c_1x7')
branch_2 = slim.conv2d(branch_2, depth(192), [7, 1],
scope='Conv2d_0d_7x1')
branch_2 = slim.conv2d(branch_2, depth(192), [1, 7],
scope='Conv2d_0e_1x7')
with tf.variable_scope('Branch_3'):
branch_3 = slim.avg_pool2d(net, [3, 3], scope='AvgPool_0a_3x3')
branch_3 = slim.conv2d(branch_3, depth(192), [1, 1],
scope='Conv2d_0b_1x1')
net = tf.concat(axis=3, values=[branch_0, branch_1, branch_2, branch_3])
end_points[end_point] = net
if end_point == final_endpoint: return net, end_points
# mixed_8: 8 x 8 x 1280.
# 之前的并聯(lián)思路， 1*1，1*1和5*5， 1*1和3*3和3*3，平均池化和1*1， 35 x 35 x 288
# 之前的并聯(lián)思路， 3*3， 1*1和3*3和3*3，最大池化，stride=2，17 x 17 x 768， 通道數(shù)增加到768
# 之前的并聯(lián)思路， 1*1，1*1和1*7和7*1， 1*1和1*7和7*1和1*7和7*1，平均池化和1*1， 17 x 17 x 768， 通道數(shù)增加到768 end_point = 'Mixed_7a'
# 這里的并聯(lián)思路， 1*1和3*3，1*1和1*7和7*1和3*3， 最大池化，8 x 8 x 1280 通道數(shù)增加到 1280
end_point = 'Mixed_7a'
with tf.variable_scope(end_point):
with tf.variable_scope('Branch_0'):
branch_0 = slim.conv2d(net, depth(192), [1, 1], scope='Conv2d_0a_1x1')
branch_0 = slim.conv2d(branch_0, depth(320), [3, 3], stride=2,
padding='VALID', scope='Conv2d_1a_3x3')
with tf.variable_scope('Branch_1'):
branch_1 = slim.conv2d(net, depth(192), [1, 1], scope='Conv2d_0a_1x1')
branch_1 = slim.conv2d(branch_1, depth(192), [1, 7],
scope='Conv2d_0b_1x7')
branch_1 = slim.conv2d(branch_1, depth(192), [7, 1],
scope='Conv2d_0c_7x1')
branch_1 = slim.conv2d(branch_1, depth(192), [3, 3], stride=2,
padding='VALID', scope='Conv2d_1a_3x3')
with tf.variable_scope('Branch_2'):
branch_2 = slim.max_pool2d(net, [3, 3], stride=2, padding='VALID',
scope='MaxPool_1a_3x3')
net = tf.concat(axis=3, values=[branch_0, branch_1, branch_2])
end_points[end_point] = net
if end_point == final_endpoint: return net, end_points
# mixed_9: 8 x 8 x 2048.
# 之前的并聯(lián)思路， 1*1，1*1和5*5， 1*1和3*3和3*3，平均池化和1*1， 35 x 35 x 288
# 之前的并聯(lián)思路， 3*3， 1*1和3*3和3*3，最大池化，stride=2，17 x 17 x 768， 通道數(shù)增加到768
# 之前的并聯(lián)思路， 1*1，1*1和1*7和7*1， 1*1和1*7和7*1和1*7和7*1，平均池化和1*1， 17 x 17 x 768， 通道數(shù)增加到768 end_point = 'Mixed_7a'
# 之前的并聯(lián)思路， 1*1和3*3，1*1和1*7和7*1和3*3， 最大池化，8 x 8 x 1280 通道數(shù)增加到 1280
# 這里的并聯(lián)思路， 1*1，1*1和1*3和3*1，這里有個 net in net 的又一次小的并聯(lián)結(jié)構(gòu)，平均池化和1*1，
# 最后得到 8 x 8 x 2048， 通道數(shù)增加到 2048
end_point = 'Mixed_7b'
with tf.variable_scope(end_point):
with tf.variable_scope('Branch_0'):
branch_0 = slim.conv2d(net, depth(320), [1, 1], scope='Conv2d_0a_1x1')
with tf.variable_scope('Branch_1'):
branch_1 = slim.conv2d(net, depth(384), [1, 1], scope='Conv2d_0a_1x1')
branch_1 = tf.concat(axis=3, values=[
slim.conv2d(branch_1, depth(384), [1, 3], scope='Conv2d_0b_1x3'),
slim.conv2d(branch_1, depth(384), [3, 1], scope='Conv2d_0b_3x1')])
with tf.variable_scope('Branch_2'):
branch_2 = slim.conv2d(net, depth(448), [1, 1], scope='Conv2d_0a_1x1')
branch_2 = slim.conv2d(
branch_2, depth(384), [3, 3], scope='Conv2d_0b_3x3')
branch_2 = tf.concat(axis=3, values=[
slim.conv2d(branch_2, depth(384), [1, 3], scope='Conv2d_0c_1x3'),
slim.conv2d(branch_2, depth(384), [3, 1], scope='Conv2d_0d_3x1')])
with tf.variable_scope('Branch_3'):
branch_3 = slim.avg_pool2d(net, [3, 3], scope='AvgPool_0a_3x3')
branch_3 = slim.conv2d(
branch_3, depth(192), [1, 1], scope='Conv2d_0b_1x1')
net = tf.concat(axis=3, values=[branch_0, branch_1, branch_2, branch_3])
end_points[end_point] = net
if end_point == final_endpoint: return net, end_points
# mixed_10: 8 x 8 x 2048. 和上面一樣
end_point = 'Mixed_7c'
with tf.variable_scope(end_point):
with tf.variable_scope('Branch_0'):
branch_0 = slim.conv2d(net, depth(320), [1, 1], scope='Conv2d_0a_1x1')
with tf.variable_scope('Branch_1'):
branch_1 = slim.conv2d(net, depth(384), [1, 1], scope='Conv2d_0a_1x1')
branch_1 = tf.concat(axis=3, values=[
slim.conv2d(branch_1, depth(384), [1, 3], scope='Conv2d_0b_1x3'),
slim.conv2d(branch_1, depth(384), [3, 1], scope='Conv2d_0c_3x1')])
with tf.variable_scope('Branch_2'):
branch_2 = slim.conv2d(net, depth(448), [1, 1], scope='Conv2d_0a_1x1')
branch_2 = slim.conv2d(
branch_2, depth(384), [3, 3], scope='Conv2d_0b_3x3')
branch_2 = tf.concat(axis=3, values=[
slim.conv2d(branch_2, depth(384), [1, 3], scope='Conv2d_0c_1x3'),
slim.conv2d(branch_2, depth(384), [3, 1], scope='Conv2d_0d_3x1')])
with tf.variable_scope('Branch_3'):
branch_3 = slim.avg_pool2d(net, [3, 3], scope='AvgPool_0a_3x3')
branch_3 = slim.conv2d(
branch_3, depth(192), [1, 1], scope='Conv2d_0b_1x1')
net = tf.concat(axis=3, values=[branch_0, branch_1, branch_2, branch_3])
end_points[end_point] = net
if end_point == final_endpoint: return net, end_points
raise ValueError('Unknown final endpoint %s' % final_endpoint)
# base 模塊最后輸出 8 x 8 x 2048
# inception_v3 結(jié)構(gòu)的主體部分
def inception_v3(inputs,
num_classes=1000,
is_training=True,
dropout_keep_prob=0.8,
min_depth=16,
depth_multiplier=1.0,
prediction_fn=slim.softmax,
spatial_squeeze=True,
reuse=None,
create_aux_logits=True,
scope='InceptionV3',
global_pool=False):
"""Inception model from http://arxiv.org/abs/1512.00567.
"Rethinking the Inception Architecture for Computer Vision"
Christian Szegedy, Vincent Vanhoucke, Sergey Ioffe, Jonathon Shlens,
Zbigniew Wojna.
With the default arguments this method constructs the exact model defined in
the paper. However, one can experiment with variations of the inception_v3
network by changing arguments dropout_keep_prob, min_depth and
depth_multiplier.
The default image size used to train this network is 299x299.
這段話意思是使用默認的參數(shù)可以得到和論文中提到的inception_v3一樣的結(jié)構(gòu)，但是你也可以通過調(diào)整
dropout_keep_prob, min_depth ，depth_multiplier 參數(shù)來得到不同的結(jié)構(gòu)變體；
另外，默認的輸入圖片尺寸為 299x299
Args:
輸入尺寸
inputs: a tensor of size [batch_size, height, width, channels].
最終輸出的分類數(shù)量
num_classes: number of predicted classes. If 0 or None, the logits layer
is omitted and the input features to the logits layer (before dropout)
are returned instead.
訓(xùn)練模式還是測試模式的開關(guān)
is_training: whether is training or not.
dropout 比例
dropout_keep_prob: the percentage of activation values that are retained.
最小通道數(shù)
min_depth: Minimum depth value (number of channels) for all convolution ops.
Enforced when depth_multiplier < 1, and not an active constraint when
depth_multiplier >= 1.
通道數(shù)倍數(shù)
depth_multiplier: Float multiplier for the depth (number of channels)
for all convolution ops. The value must be greater than zero. Typical
usage will be to set this value in (0, 1) to reduce the number of
parameters or computation cost of the model.
logits 之后的輸出經(jīng)過的函數(shù)，默認為 softmax
prediction_fn: a function to get predictions out of logits.
是否壓縮去掉數(shù)目為1的維度
spatial_squeeze: if True, logits is of shape [B, C], if false logits is of
shape [B, 1, 1, C], where B is batch_size and C is number of classes.
變量是否重復(fù)使用
reuse: whether or not the network and its variables should be reused. To be
able to reuse 'scope' must be given.
是否創(chuàng)建輔助logits 節(jié)點
create_aux_logits: Whether to create the auxiliary logits.
scope: Optional variable_scope.
是否使用全局池化，默認不使用，使用的情況下，任意大小的輸入都會被平均，導(dǎo)致輸出為1*1
global_pool: Optional boolean flag to control the avgpooling before the
logits layer. If false or unset, pooling is done with a fixed window
that reduces default-sized inputs to 1x1, while larger inputs lead to
larger outputs. If true, any input size is pooled down to 1x1.
Returns:
輸出，logits
net: a Tensor with the logits (pre-softmax activations) if num_classes
is a non-zero integer, or the non-dropped-out input to the logits layer
if num_classes is 0 or None.
網(wǎng)絡(luò)各個節(jié)點的數(shù)據(jù)
end_points: a dictionary from components of the network to the corresponding
activation.
Raises:
ValueError: if 'depth_multiplier' is less than or equal to zero.
"""
if depth_multiplier <= 0:
raise ValueError('depth_multiplier is not greater than zero.')
depth = lambda d: max(int(d * depth_multiplier), min_depth)
# 這里開頭就使用 base 模塊構(gòu)建了 inception 網(wǎng)絡(luò)的前面部分結(jié)構(gòu)，在輔助節(jié)點之后又構(gòu)建了網(wǎng)絡(luò)的尾部
# 構(gòu)建了網(wǎng)絡(luò)的 arg_scope， 一個問題是這里都沒有用到 get_variable， 那這個 reuse 參數(shù)的影響又體現(xiàn)在哪里呢？
with tf.variable_scope(scope, 'InceptionV3', [inputs], reuse=reuse) as scope:
# 設(shè)置 slim.batch_norm, slim.dropout 的參數(shù)值， dropout 倒是只在最后的輸出前面一層用到了一次，
# 而 batch_norm 則是在每個卷積層都默認使用了，因為事先指定了slim.conv2d 的 normalizer_fn 參數(shù)
with slim.arg_scope([slim.batch_norm, slim.dropout],
is_training=is_training):
net, end_points = inception_v3_base(
inputs, scope=scope, min_depth=min_depth,
depth_multiplier=depth_multiplier)
# Auxiliary Head logits 輔助節(jié)點信息，是從 Mixed_6e 拉出來的節(jié)點
if create_aux_logits and num_classes:
with slim.arg_scope([slim.conv2d, slim.max_pool2d, slim.avg_pool2d],
stride=1, padding='SAME'):
aux_logits = end_points['Mixed_6e']
with tf.variable_scope('AuxLogits'):
aux_logits = slim.avg_pool2d(
aux_logits, [5, 5], stride=3, padding='VALID',
scope='AvgPool_1a_5x5')
aux_logits = slim.conv2d(aux_logits, depth(128), [1, 1],
scope='Conv2d_1b_1x1')
# Shape of feature map before the final layer.
kernel_size = _reduced_kernel_size_for_small_input(
aux_logits, [5, 5])
aux_logits = slim.conv2d(
aux_logits, depth(768), kernel_size,
weights_initializer=trunc_normal(0.01),
padding='VALID', scope='Conv2d_2a_{}x{}'.format(*kernel_size))
aux_logits = slim.conv2d(
aux_logits, num_classes, [1, 1], activation_fn=None,
normalizer_fn=None, weights_initializer=trunc_normal(0.001),
scope='Conv2d_2b_1x1')
if spatial_squeeze:
aux_logits = tf.squeeze(aux_logits, [1, 2], name='SpatialSqueeze')
end_points['AuxLogits'] = aux_logits
# Final pooling and prediction 網(wǎng)絡(luò)的尾部，base 輸出為 8 x 8 x 2048，經(jīng)過全局平均池化后為 1 x 1 x 2048
with tf.variable_scope('Logits'):
if global_pool:
# Global average pooling.
net = tf.reduce_mean(net, [1, 2], keep_dims=True, name='GlobalPool')
end_points['global_pool'] = net
else:
# Pooling with a fixed kernel size.
# 這里 global_pool 為 T 和 F 效果都是一樣的，因為輸入 是8 x 8 x 2048， 而這里的平均池化的核大小也是 8*8
kernel_size = _reduced_kernel_size_for_small_input(net, [8, 8])
net = slim.avg_pool2d(net, kernel_size, padding='VALID',
scope='AvgPool_1a_{}x{}'.format(*kernel_size))
end_points['AvgPool_1a'] = net
if not num_classes:
return net, end_points
# 1 x 1 x 2048
net = slim.dropout(net, keep_prob=dropout_keep_prob, scope='Dropout_1b')
end_points['PreLogits'] = net
# 2048
# 這里得到分類結(jié)果，由1 x 1 x 2048變?yōu)?【batch，1，1，num_classes】
logits = slim.conv2d(net, num_classes, [1, 1], activation_fn=None,
normalizer_fn=None, scope='Conv2d_1c_1x1')
if spatial_squeeze: # 壓縮去掉 1和2 維度
logits = tf.squeeze(logits, [1, 2], name='SpatialSqueeze')
# 1000
end_points['Logits'] = logits
end_points['Predictions'] = prediction_fn(logits, scope='Predictions')
return logits, end_points
inception_v3.default_image_size = 299
def _reduced_kernel_size_for_small_input(input_tensor, kernel_size):
"""
定義網(wǎng)絡(luò)的卷積核的大小，使它不超過輸入的 tensor 的大小，
在輸入的 tensor 尺寸較小的時候發(fā)揮作用，
Define kernel size which is automatically reduced for small input.
If the shape of the input images is unknown at graph construction time this
function assumes that the input images are is large enough.
Args:
input_tensor: input tensor of size [batch_size, height, width, channels].
kernel_size: desired kernel size of length 2: [kernel_height, kernel_width]
Returns:
a tensor with the kernel size.
TODO(jrru): Make this function work with unknown shapes. Theoretically, this
can be done with the code below. Problems are two-fold: (1) If the shape was
known, it will be lost. (2) inception.slim.ops._two_element_tuple cannot
handle tensors that define the kernel size.
shape = tf.shape(input_tensor)
return = tf.stack([tf.minimum(shape[1], kernel_size[0]),
tf.minimum(shape[2], kernel_size[1])])
"""
shape = input_tensor.get_shape().as_list()
if shape[1] is None or shape[2] is None:
kernel_size_out = kernel_size
else:
kernel_size_out = [min(shape[1], kernel_size[0]),
min(shape[2], kernel_size[1])]
return kernel_size_out
# 這里倒是有點疑問， 在 inception_v3 的主體部分已經(jīng)設(shè)置了 arg_scope ，為什么這里又要再設(shè)置一次
inception_v3_arg_scope = inception_utils.inception_arg_scope
# 上面用到的 inception_arg_scope 是一個如下的函數(shù)，它返回整個 inception 網(wǎng)絡(luò)的參數(shù)作用域
def inception_arg_scope(weight_decay=0.00004,
use_batch_norm=True,
batch_norm_decay=0.9997,
batch_norm_epsilon=0.001,
activation_fn=tf.nn.relu):
"""Defines the default arg scope for inception models.
Args:
weight_decay: The weight decay to use for regularizing the model.
use_batch_norm: "If `True`, batch_norm is applied after each convolution.
batch_norm_decay: Decay for batch norm moving average.
batch_norm_epsilon: Small float added to variance to avoid dividing by zero
in batch norm.
activation_fn: Activation function for conv2d.
Returns:
An `arg_scope` to use for the inception models.
"""
batch_norm_params = {
# Decay for the moving averages.
'decay': batch_norm_decay,
# epsilon to prevent 0s in variance.
'epsilon': batch_norm_epsilon,
# collection containing update_ops.
'updates_collections': tf.GraphKeys.UPDATE_OPS,
# use fused batch norm if possible.
'fused': None,
}
# 是否使用BN批標準化操作
if use_batch_norm:
normalizer_fn = slim.batch_norm
normalizer_params = batch_norm_params
else:
normalizer_fn = None
normalizer_params = {}
# Set weight_decay for weights in Conv and FC layers.
# 嵌套的給卷積層和全連接層設(shè)置參數(shù)正則化；
# 嵌套的給卷積層設(shè)置參數(shù)初始化、激活函數(shù)、BN批標準化
with slim.arg_scope([slim.conv2d, slim.fully_connected],
weights_regularizer=slim.l2_regularizer(weight_decay)):
with slim.arg_scope(
[slim.conv2d],
weights_initializer=slim.variance_scaling_initializer(),
activation_fn=activation_fn,
normalizer_fn=normalizer_fn,
normalizer_params=normalizer_params) as sc:
return sc通讀下來，發(fā)現(xiàn)它的結(jié)構(gòu)理解起來還是比較簡單的質(zhì)量效應(yīng)3男主默認代碼，我又找了兩張圖，可以更加好的從整體把握它的結(jié)構(gòu)：
的整體結(jié)構(gòu)，是一個多個并聯(lián)模塊串聯(lián)的結(jié)構(gòu)，最后兩個模塊使用了更加復(fù)雜的結(jié)構(gòu)，從 拉出來輔助節(jié)點
其中某個并聯(lián)模塊的結(jié)構(gòu)
代碼鏈接：