学习了Inception V3卷积神经网络，总结一下对Inception V3网络结构和主要代码的理解。
GoogLeNet对网络中的传统卷积层进行了修改，提出了被称为 Inception 的结构，用于增加网络深度和宽度，提高深度神经网络性能。从Inception V1到Inception V4有4个更新版本，每一版的网络在原来的基础上进行改进，提高网络性能。本文介绍Inception V3的网络结构和主要代码。
首先定义一个非Inception Module的普通卷积层函数inception_v3_base，输入参数inputs为图片数据的张量。第1个卷积层的输出通道数为32，卷积核尺寸为【3x3】，步长为2，padding模式是默认的VALID，第1个卷积层之后的张量尺寸变为(299-3)/2+1=149，即【149x149x32】。
后面的卷积层采用相同的形式，最后张量尺寸变为【35x35x192】。这几个普通的卷积层主要使用了3x3的小卷积核，小卷积核可以低成本的跨通道的对特征进行组合。
def inception_v3_base(inputs,scepe=None):
with tf.variable_scope(scope,'InceptionV3',[inputs]):
with slim.arg_scope([slim.conv2d,slim.max_pool2d,slim.avg_pool2d],stride=1,padding='VALID'):
# 149 x 149 x 32
net = slim.conv2d(inputs,32,[3,3],stride=2,scope='Conv2d_1a_3x3')
# 147 x 147 x 32'
net = slim.conv2d(net,32),[3,3],scope='Conv2d_2a_3x3')
# 147 x 147 x 64
net = slim.conv2d(net,64,[3,3],padding='SAME',scope='Conv2d_2b_3x3')
# 73 x 73 x 64
net = slim.max_pool2d(net, [3, 3], stride=2, scope='MaxPool_3a_3x3')
# 73 x 73 x 80
net = slim.conv2d(net, 80, [1, 1], scope= 'Conv2d_3b_1x1')
# 71 x 71 x 192.
net = slim.conv2d(net, 192, [3, 3], scope='Conv2d_4a_3x3',reuse=tf.AUTO_REUSE)
# 35 x 35 x 192
net = slim.max_pool2d(net, [3, 3], stride=2, scope= 'MaxPool_5a_3x3') 接下来是三个连续的Inception模块组，每个模块组有多个Inception module组成。
下面是第1个Inception模块组，包含了3个类似的Inception module，分别是：Mixed_5b，Mixed_5c，Mixed_5d。第1个Inception module有4个分支，
第1个分支是输出通道为64的【1x1】卷积，
第2个分支是输出通道为48的【1x1】卷积，再连接输出通道为64的【5x5】卷积，
第3个分支是输出通道为64的【1x1】卷积，再连接2个输出通道为96的【3x3】卷积，
第4个分支是【3x3】的平均池化，再连接输出通道为32的【1x1】卷积。
最后用tf.concat将4个分支的输出合并在一起，输出通道之和为54+64+96+32=256，最后输出的张量尺寸为【35x35x256】。
第2个Inception module也有4个分支，与第1个模块类似，只是最后连接输出通道数为64的【1x1】卷积，最后输出的张量尺寸为【35x35x288】。
第3个模块与第2个模块一样。
with slim.arg_scope([slim.conv2d,slim.max_pool2d,slim.avg_pool2d],stride=1,padding='SAME'):
# 35 x 35 x 256
end_point = 'Mixed_5b'
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(32), [1, 1], scope='Conv2d_0b_1x1')
net = tf.concat(axis=3, values=[branch_0, branch_1, branch_2, branch_3]) # 64+64+96+32=256
end_points[end_point] = net
# 35 x 35 x 288
end_point = 'Mixed_5c'
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_0b_1x1')
branch_1 = slim.conv2d(branch_1, depth(64), [5, 5],scope='Conv_1_0c_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
# 35 x 35 x 288
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第2个Inception模块组包含了5个Inception module，分别是Mixed_6a，Mixed_6b，Mixed_6ac，Mixed_6d，Mixed_6e。
每个Inception module包含有多个分支，第1个Inception module的步长为2，因此图片尺寸被压缩，最后输出的张量尺寸为【17x17x768】。
第2个Inception module采用了Fractorization into small convolutions思想，串联了【1x7】和【7x1】卷积，最后也是将多个通道合并。
第3、4个Inception module与第2个类似，都是用来增加卷积和非线性变化，提炼特征。张量尺寸不变，多个module后仍旧是【17x17x768】。
# 17 x 17 x 768.
end_point = 'Mixed_6a'
with tf.variable_scope(end_point):
with tf.variable_scope('Branch_0'):
branch_0 = slim.conv2d(net, depth(384), [3, 3], stride=2,padding='VALID', scope='Conv2d_1a_1x1')
with tf.variable_scope('Branch_1'):
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'):
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]) # (35-3)/2+1=17
end_points[end_point] = net
# 17 x 17 x 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
print(net.shape)
# 17 x 17 x 768.
end_point = 'Mixed_6c'
with tf.variable_scope(end_point):
with tf.variable_scope('Branch_0'):
ranch_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
# 17 x 17 x 768.
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], sco e='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
# 17 x 17 x 768.
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第3个Inception模块组包含了3个Inception module，分别是Mxied_7a，Mixed_7b，Mixed_7c。
第1个Inception module包含了3个分支，与上面的结构类似，主要也是通过改变通道数、卷积核尺寸，包括【1x1】、【3x3】、【1x7】、【7x1】来增加卷积和非线性变化，提升网络性能。
最后3个分支在输出通道上合并，输出张量的尺寸为【8 x 8 x 1280】。第3个Inception module后得到的张量尺寸为【8 x 8 x 2048】。
# 8 x 8 x 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
# 8 x 8 x 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)
# 8 x 8 x 2048.
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] = netInception V3网络的最后一部分是Auxiliary Logits、全局平均池化、Softmax分类。
首先是Auxiliary Logits，作为辅助分类的节点，对分类结果预测有很大帮助。
先通过end_points['Mixed_6e']得到Mixed_6e后的特征张量，之后接一个【5x5】的平均池化，步长为3，padding为VALID，张量尺寸从第2个模块组的【17x17x768】变为【5x5x768】。
接着连接一个输出通道为128的【1x1】卷积和输出通道为768的【5x5】卷积，输出尺寸变为【1x1x768】。
然后连接输出通道数为num_classes的【1x1】卷积，输出变为【1x1x1000】。最后将辅助分类节点的输出存储到字典表end_points中。
with slim.arg_scope([slim.conv2d,slim.max_pool2d,slim.avg_pool2d],stride=1,padding='SAME'):
aux_logits = end_points['Mixed_6e']
print(aux_logits.shape)
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') # (17-5)/3+1=5
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')
aux_logits = tf.squeeze(aux_logits, [1, 2], name='SpatialSqueeze')
end_points['AuxLogits'] = aux_logits最后对最后一个卷积层的输出Mixed_7c进行一个【8x8】的全局平均池化，padding为VALID，输出张量从【8 x 8 x 2048】变为【1 x 1 x 2048】，然后连接一个Dropout层，接着连接一个输出通道数为1000的【1x1】卷积。
使用tf.squeeze去掉输出张量中维数为1的维度。最后用Softmax得到最终分类结果。返回分类结果logits和包含各个卷积后的特征图字典表end_points。
with tf.variable_scope('Logits'):
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
net = slim.dropout(net, keep_prob=dropout_keep_prob, scope='Dropout_1b')
end_points['PreLogits'] = net
logits = slim.conv2d(net, num_classes, [1, 1], activation_fn=None, normalizer_fn=None, scope='Conv2d_1c_1x1')
logits = tf.squeeze(logits, [1, 2], name='SpatialSqueeze')
end_points['Logits'] = logits
end_points['Predictions'] = slim.softmax(logits, scope='Predictions')
return logits,end_points参考文献：
1. 《TensorFlow实战》
