keras的三种模型实现与区别说明

阳光穿透心脏的1/2处 2022-12-26 00:58 258阅读 0赞

更多编程教程请到：[菜鸟教程][Link 1] https://www.piaodoo.com/

友情链接：

[高州阳光论坛][Link 2]https://www.hnthzk.com/

[人人影视][Link 3]http://www.op-kg.com/

**前言**

一、keras提供了三种定义模型的方式

**1. 序列式(Sequential) API**

序贯(sequential)API允许你为大多数问题逐层堆叠创建模型。虽然说对很多的应用来说，这样的一个手法很简单也解决了很多深度学习网络结构的构建，但是它也有限制－它不允许你创建模型有共享层或有多个输入或输出的网络。

**2. 函数式(Functional) API**

Keras函数式(functional)API为构建网络模型提供了更为灵活的方式。

它允许你定义多个输入或输出模型以及共享图层的模型。除此之外，它允许你定义动态(ad-hoc)的非周期性(acyclic)网络图。

模型是通过创建层的实例(layer instances)并将它们直接相互连接成对来定义的，然后定义一个模型(model)来指定那些层是要作为这个模型的输入和输出。

**3.子类(Subclassing) API**

**补充知识：****keras pytorch 构建模型对比**

使用CIFAR10数据集，用三种框架构建Residual\_Network作为例子，比较框架间的异同。

**数据集格式**

pytorch的数据集格式

import torch
    import torch.nn as nn
    import torchvision

# Download and construct CIFAR-10 dataset. #

train\_dataset = torchvision.datasets.CIFAR10(root=’…/…/data/’,  
train=True,  
download=True)

# Fetch one data pair (read data from disk). #

image, label = train\_dataset\[0\]  
print (image.size()) \# torch.Size(\[3, 32, 32\])  
print (label) \# 6  
print (train\_dataset.data.shape) \# (50000, 32, 32, 3)

# type(train\_dataset.targets)==list #

print (len(train\_dataset.targets)) \# 50000

# Data loader (this provides queues and threads in a very simple way). #

train\_loader = torch.utils.data.DataLoader(dataset=train\_dataset,  
batch\_size=64,  
shuffle=True)  
“”"

# 演示DataLoader返回的数据结构 #

# When iteration starts, queue and thread start to load data from files. #

data\_iter = iter(train\_loader)

# Mini-batch images and labels. #

images, labels = data\_iter.next()  
print(images.shape) \# torch.Size(\[100, 3, 32, 32\])  
print(labels.shape)

# torch.Size(\[100\]) 可见经过DataLoader后，labels由list变成了pytorch内置的tensor格式 #

“”"

# 一般使用的话是下面这种 #

# Actual usage of the data loader is as below. #

for images, labels in train\_loader:

# Training code should be written here. #

pass

**keras的数据格式**

import keras
    from keras.datasets import cifar10

(train\_x, train\_y) , (test\_x, test\_y) = cifar10.load\_data()  
print(train\_x.shape) \# ndarray 类型： (50000, 32, 32, 3)  
print(train\_y.shape) \# (50000, 1)

**输入网络的数据格式不同**

"""
    1: pytorch 都是内置torch.xxTensor输入网络，而keras的则是原生ndarray类型
    2: 对于multi-class的其中一种loss，即cross-entropy loss 而言，
      pytorch的api为 CorssEntropyLoss, 但y_true不能用one-hoe编码！这与keras，tensorflow	    都不同。tensorflow相应的api为softmax_cross_entropy
      他们的api都仅限于multi-class classification
    3*: 其实上面提到的api都属于categorical cross-entropy loss,
      又叫 softmax loss，是函数内部先进行了 softmax 激活，再经过cross-entropy loss。
      这个loss是cross-entropy loss的变种，
      cross-entropy loss又叫logistic loss 或 multinomial logistic loss。
      实现这种loss的函数不包括激活函数，需要自定义。
      pytorch对应的api为BCEloss(仅限于 binary classification),
      tensorflow 对应的api为 log_loss。
      cross-entropy loss的第二个变种是 binary cross-entropy loss 又叫 sigmoid cross-  entropy loss。
      函数内部先进行了sigmoid激活，再经过cross-entropy loss。
      pytorch对应的api为BCEWithLogitsLoss,
      tensorflow对应的api为sigmoid_cross_entropy
    """

# pytorch #

criterion = nn.CrossEntropyLoss()  
…  
for epoch in range(num\_epochs):  
for i, (images, labels) in enumerate(train\_loader):  
images = images.to(device)  
labels = labels.to(device)

# Forward pass
    outputs = model(images)
    # 对于multi-class cross-entropy loss
    # 输入y_true不需要one-hot编码
    loss = criterion(outputs, labels)

…

# keras #

# 对于multi-class cross-entropy loss #

# 输入y\_true需要one-hot编码 #

train\_y = keras.utils.to\_categorical(train\_y,10)  
…  
model.fit\_generator(datagen.flow(train\_x, train\_y, batch\_size=128),  
validation\_data=\[test\_x,test\_y\],  
epochs=epochs,steps\_per\_epoch=steps\_per\_epoch, verbose=1)  
…

**整体流程**

**keras 流程**

model = myModel()
    model.compile(optimizer=Adam(0.001),loss="categorical_crossentropy",metrics=["accuracy"])
    model.fit_generator(datagen.flow(train_x, train_y, batch_size=128),
              validation_data=[test_x,test_y],
              epochs=epochs,steps_per_epoch=steps_per_epoch, verbose=1, workers=4)
    #Evaluate the accuracy of the test dataset
    accuracy = model.evaluate(x=test_x,y=test_y,batch_size=128)
    # 保存整个网络
    model.save("cifar10model.h5")
    """
    # https://blog.csdn.net/jiandanjinxin/article/details/77152530
    # 使用
    # keras.models.load_model("cifar10model.h5")

# 只保存architecture #

# json\_string = model.to\_json() #

# open(‘my\_model\_architecture.json’,‘w’).write(json\_string) #

# 使用 #

# from keras.models import model\_from\_json #

\#model = model\_from\_json(open(‘my\_model\_architecture.json’).read())

# 只保存weights #

# model.save\_weights(‘my\_model\_weights.h5’) #

\#需要在代码中初始化一个完全相同的模型

# model.load\_weights(‘my\_model\_weights.h5’) #

\#需要加载权重到不同的网络结构（有些层一样）中，例如fine-tune或transfer-learning，可以通过层名字来加载模型

# model.load\_weights(‘my\_model\_weights.h5’, by\_name=True) #

“”"

**pytorch 流程**

model = myModel()
    # Loss and optimizer
    criterion = nn.CrossEntropyLoss()

for epoch in range(num\_epochs):  
for i, (images, labels) in enumerate(train\_loader):  
images = images.to(device)  
labels = labels.to(device)

# Forward pass
    outputs = model(images)
    loss = criterion(outputs, labels)
    
    # Backward and optimize
    	# 将上次迭代计算的梯度值清0
    optimizer.zero_grad()
    # 反向传播，计算梯度值
    loss.backward()
    # 更新权值参数
    optimizer.step()

# model.eval()，让model变成测试模式，对dropout和batch normalization的操作在训练和测试的时候是不一样的 #

# eval（）时，pytorch会自动把BN和DropOut固定住，不会取平均，而是用训练好的值。 #

# 不然的话，一旦test的batch\_size过小，很容易就会被BN层导致生成图片颜色失真极大。 #

model.eval()  
with torch.no\_grad():  
correct = 0  
total = 0  
for images, labels in test\_loader:  
images = images.to(device)  
labels = labels.to(device)  
outputs = model(images)  
\_, predicted = torch.max(outputs.data, 1)  
total += labels.size(0)  
correct += (predicted == labels).sum().item()

print(‘Accuracy of the model on the test images: \{\} %’.format(100 \* correct / total))

# Save the model checkpoint #

# 这是只保存了weights #

torch.save(model.state\_dict(), ‘resnet.ckpt’)  
“”"

# 使用 #

# myModel.load\_state\_dict(torch.load(‘params.ckpt’)) #

# 若想保存整个网络（architecture + weights) #

# torch.save(resnet, ‘model.ckpt’) #

# 使用 #

\#model = torch.load(‘model.ckpt’)  
“”"

**对比流程**

#https://blog.csdn.net/dss_dssssd/article/details/83892824
    """
    1: 准备数据(注意数据格式不同）
    2: 定义网络结构model
    3: 定义损失函数
    4: 定义优化算法 optimizer
    5: 训练-keras
    	5.1:编译模型（传入loss function和optimizer等）
    	5.2:训练模型（fit or fit_generator，传入数据)
    5: 训练-pytorch
    迭代训练：
    	5.1:准备好tensor形式的输入数据和标签(可选)
    	5.2:前向传播计算网络输出output和计算损失函数loss
    	5.3:反向传播更新参数
    		以下三句话一句也不能少：
    		5.3.1:将上次迭代计算的梯度值清0
    			optimizer.zero_grad()
    		5.3.2:反向传播，计算梯度值
    			loss.backward()
    		5.3.3:更新权值参数
    			optimizer.step()
    6: 在测试集上测试-keras
    	model.evaluate
    6: 在测试集上测试-pytorch
      遍历测试集，自定义metric
    7: 保存网络（可选） 具体实现参考上面代码
    """

**构建网络**

**对比网络**

1、对于keras，不需要input\_channels，函数内部会自动获得，而pytorch则需要显示声明input\_channels

2、对于pytorch Conv2d需要指定padding，而keras的则是same和valid两种选项（valid即padding=0）

3、keras的Flatten操作可以视作pytorch中的view

4、keras的dimension一般顺序是（H, W, C) (tensorflow 为backend的话），而pytorch的顺序则是（ C, H, W)

5、具体的变换可以参照下方，但由于没有学过pytorch，keras也刚入门，不能保证正确，日后学的更深入了之后再来看看。

**pytorch 构建Residual-network**

import torch
    import torch.nn as nn
    import torchvision
    import torchvision.transforms as transforms

# Device configuration #

device = torch.device(‘cuda’ if torch.cuda.is\_available() else ‘cpu’)

# Hyper-parameters #

num\_epochs = 80  
learning\_rate = 0.001

# Image preprocessing modules #

transform = transforms.Compose(\[  
transforms.Pad(4),  
transforms.RandomHorizontalFlip(),  
transforms.RandomCrop(32),  
transforms.ToTensor()\])

# CIFAR-10 dataset #

# train\_dataset.data.shape #

\#Out\[31\]: (50000, 32, 32, 3)

# train\_dataset.targets list #

# len(list)=5000 #

train\_dataset = torchvision.datasets.CIFAR10(root=’./data/’,  
train=True,  
transform=transform,  
download=True)

test\_dataset = torchvision.datasets.CIFAR10(root=’…/…/data/’,  
train=False,  
transform=transforms.ToTensor())

# Data loader #

train\_loader = torch.utils.data.DataLoader(dataset=train\_dataset,  
batch\_size=100,  
shuffle=True)

test\_loader = torch.utils.data.DataLoader(dataset=test\_dataset,  
batch\_size=100,  
shuffle=False)

# 3x3 convolution #

def conv3x3(in\_channels, out\_channels, stride=1):  
return nn.Conv2d(in\_channels, out\_channels, kernel\_size=3,  
stride=stride, padding=1, bias=False)

# Residual block #

class ResidualBlock(nn.Module):  
def **init**(self, in\_channels, out\_channels, stride=1, downsample=None):  
super(ResidualBlock, self).**init**()  
self.conv1 = conv3x3(in\_channels, out\_channels, stride)  
self.bn1 = nn.BatchNorm2d(out\_channels)  
self.relu = nn.ReLU(inplace=True)  
self.conv2 = conv3x3(out\_channels, out\_channels)  
self.bn2 = nn.BatchNorm2d(out\_channels)  
self.downsample = downsample

# ResNet #

class ResNet(nn.Module):  
def **init**(self, block, layers, num\_classes=10):  
super(ResNet, self).**init**()  
self.in\_channels = 16  
self.conv = conv3x3(3, 16)  
self.bn = nn.BatchNorm2d(16)  
self.relu = nn.ReLU(inplace=True)  
self.layer1 = self.make\_layer(block, 16, layers\[0\])  
self.layer2 = self.make\_layer(block, 32, layers\[1\], 2)  
self.layer3 = self.make\_layer(block, 64, layers\[2\], 2)  
self.avg\_pool = nn.AvgPool2d(8)  
self.fc = nn.Linear(64, num\_classes)

def make\_layer(self, block, out\_channels, blocks, stride=1):  
downsample = None  
if (stride != 1) or (self.in\_channels != out\_channels):  
downsample = nn.Sequential(  
conv3x3(self.in\_channels, out\_channels, stride=stride),  
nn.BatchNorm2d(out\_channels))  
layers = \[\]  
layers.append(block(self.in\_channels, out\_channels, stride, downsample))  
self.in\_channels = out\_channels  
for i in range(1, blocks):  
layers.append(block(out\_channels, out\_channels))  
\# \[\*\[1,2,3\]\]  
\# Out\[96\]: \[1, 2, 3\]  
return nn.Sequential(\*layers)

def forward(self, x):  
out = self.conv(x) \# out.shape:torch.Size(\[100, 16, 32, 32\])  
out = self.bn(out)  
out = self.relu(out)  
out = self.layer1(out)  
out = self.layer2(out)  
out = self.layer3(out)  
out = self.avg\_pool(out)  
out = out.view(out.size(0), -1)  
out = self.fc(out)  
return out

model = ResNet(ResidualBlock, \[2, 2, 2\]).to(device)

# pip install torchsummary or #

# git clone https://github.com/sksq96/pytorch-summary #

from torchsummary import summary

# input\_size=(C,H,W) #

summary(model, input\_size=(3, 32, 32))

images,labels = iter(train\_loader).next()  
outputs = model(images)

# Loss and optimizer #

criterion = nn.CrossEntropyLoss()  
optimizer = torch.optim.Adam(model.parameters(), lr=learning\_rate)

# For updating learning rate #

def update\_lr(optimizer, lr):  
for param\_group in optimizer.param\_groups:  
param\_group\[‘lr’\] = lr

# Train the model #

total\_step = len(train\_loader)  
curr\_lr = learning\_rate  
for epoch in range(num\_epochs):  
for i, (images, labels) in enumerate(train\_loader):  
images = images.to(device)  
labels = labels.to(device)

# Forward pass
    outputs = model(images)
    loss = criterion(outputs, labels)
    
    # Backward and optimize
    optimizer.zero_grad()
    loss.backward()
    optimizer.step()
    
    if (i+1) % 100 == 0:
      print ("Epoch [{}/{}], Step [{}/{}] Loss: {:.4f}"
          .format(epoch+1, num_epochs, i+1, total_step, loss.item()))

# Decay learning rate #

if (epoch+1) % 20 == 0:  
curr\_lr /= 3  
update\_lr(optimizer, curr\_lr)

# Test the model #

print(‘Accuracy of the model on the test images: \{\} %’.format(100 \* correct / total))

# Save the model checkpoint #

torch.save(model.state\_dict(), ‘resnet.ckpt’)

keras 对应的网络构建部分

"""
    #pytorch
    def conv3x3(in_channels, out_channels, stride=1):
      return nn.Conv2d(in_channels, out_channels, kernel_size=3, 
               stride=stride, padding=1, bias=False)
    """

def conv3x3(x,out\_channels, stride=1):  
\#out = spatial\_2d\_padding(x,padding=((1, 1), (1, 1)), data\_format=“channels\_last”)  
return Conv2D(filters=out\_channels, kernel\_size=\[3,3\], strides=(stride,stride),padding=“same”)(x)

“”"

# pytorch #

# Residual block #

def forward(self, x):  
residual = x  
out = self.conv1(x)  
out = self.bn1(out)  
out = self.relu(out)  
out = self.conv2(out)  
out = self.bn2(out)  
if self.downsample:  
residual = self.downsample(x)  
out += residual  
out = self.relu(out)  
return out  
“”"  
def ResidualBlock(x, out\_channels, stride=1, downsample=False):  
residual = x  
out = conv3x3(x, out\_channels,stride)  
out = BatchNormalization()(out)  
out = Activation(“relu”)(out)  
out = conv3x3(out, out\_channels)  
out = BatchNormalization()(out)  
if downsample:  
residual = conv3x3(residual, out\_channels, stride=stride)  
residual = BatchNormalization()(residual)  
out = keras.layers.add(\[residual,out\])  
out = Activation(“relu”)(out)  
return out  
“”"  
\#pytorch  
def make\_layer(self, block, out\_channels, blocks, stride=1):  
downsample = None  
if (stride != 1) or (self.in\_channels != out\_channels):  
downsample = nn.Sequential(  
conv3x3(self.in\_channels, out\_channels, stride=stride),  
nn.BatchNorm2d(out\_channels))  
layers = \[\]  
layers.append(block(self.in\_channels, out\_channels, stride, downsample))  
self.in\_channels = out\_channels  
for i in range(1, blocks):  
layers.append(block(out\_channels, out\_channels))  
\# \[\*\[1,2,3\]\]  
\# Out\[96\]: \[1, 2, 3\]  
return nn.Sequential(\*layers)  
“”"  
def make\_layer(x, out\_channels, blocks, stride=1):  
\# tf backend: x.output\_shape\[-1\]==out\_channels  
\#print("x.shape\[-1\] ",x.shape\[-1\])  
downsample = False  
if (stride != 1) or (out\_channels != x.shape\[-1\]):  
downsample = True  
out = ResidualBlock(x, out\_channels, stride, downsample)  
for i in range(1, blocks):  
out = ResidualBlock(out, out\_channels)  
return out

def KerasResidual(input\_shape):  
images = Input(input\_shape)  
out = conv3x3(images,16) \# out.shape=(None, 32, 32, 16)  
out = BatchNormalization()(out)  
out = Activation(“relu”)(out)  
layer1\_out = make\_layer(out, 16, layers\[0\])  
layer2\_out = make\_layer(layer1\_out, 32, layers\[1\], 2)  
layer3\_out = make\_layer(layer2\_out, 64, layers\[2\], 2)  
out = AveragePooling2D(pool\_size=(8,8))(layer3\_out)  
out = Flatten()(out)

# pytorch 的nn.CrossEntropyLoss()会首先执行softmax计算 #

# 当换成keras时，没有tf类似的softmax\_cross\_entropy #

# 自带的categorical\_crossentropy不会执行激活操作，因此得在Dense层加上activation #

out = Dense(units=10, activation=“softmax”)(out)  
model = Model(inputs=images,outputs=out)  
return model

input\_shape=(32, 32, 3)  
layers=\[2, 2, 2\]  
mymodel = KerasResidual(input\_shape)  
mymodel.summary()

pytorch model summary

----------------------------------------------------------------
        Layer (type)        Output Shape     Param #
    ================================================================
          Conv2d-1      [-1, 16, 32, 32]       432
        BatchNorm2d-2      [-1, 16, 32, 32]       32
           ReLU-3      [-1, 16, 32, 32]        0
          Conv2d-4      [-1, 16, 32, 32]      2,304
        BatchNorm2d-5      [-1, 16, 32, 32]       32
           ReLU-6      [-1, 16, 32, 32]        0
          Conv2d-7      [-1, 16, 32, 32]      2,304
        BatchNorm2d-8      [-1, 16, 32, 32]       32
           ReLU-9      [-1, 16, 32, 32]        0
      ResidualBlock-10      [-1, 16, 32, 32]        0
          Conv2d-11      [-1, 16, 32, 32]      2,304
       BatchNorm2d-12      [-1, 16, 32, 32]       32
           ReLU-13      [-1, 16, 32, 32]        0
          Conv2d-14      [-1, 16, 32, 32]      2,304
       BatchNorm2d-15      [-1, 16, 32, 32]       32
           ReLU-16      [-1, 16, 32, 32]        0
      ResidualBlock-17      [-1, 16, 32, 32]        0
          Conv2d-18      [-1, 32, 16, 16]      4,608
       BatchNorm2d-19      [-1, 32, 16, 16]       64
           ReLU-20      [-1, 32, 16, 16]        0
          Conv2d-21      [-1, 32, 16, 16]      9,216
       BatchNorm2d-22      [-1, 32, 16, 16]       64
          Conv2d-23      [-1, 32, 16, 16]      4,608
       BatchNorm2d-24      [-1, 32, 16, 16]       64
           ReLU-25      [-1, 32, 16, 16]        0
      ResidualBlock-26      [-1, 32, 16, 16]        0
          Conv2d-27      [-1, 32, 16, 16]      9,216
       BatchNorm2d-28      [-1, 32, 16, 16]       64
           ReLU-29      [-1, 32, 16, 16]        0
          Conv2d-30      [-1, 32, 16, 16]      9,216
       BatchNorm2d-31      [-1, 32, 16, 16]       64
           ReLU-32      [-1, 32, 16, 16]        0
      ResidualBlock-33      [-1, 32, 16, 16]        0
          Conv2d-34       [-1, 64, 8, 8]     18,432
       BatchNorm2d-35       [-1, 64, 8, 8]       128
           ReLU-36       [-1, 64, 8, 8]        0
          Conv2d-37       [-1, 64, 8, 8]     36,864
       BatchNorm2d-38       [-1, 64, 8, 8]       128
          Conv2d-39       [-1, 64, 8, 8]     18,432
       BatchNorm2d-40       [-1, 64, 8, 8]       128
           ReLU-41       [-1, 64, 8, 8]        0
      ResidualBlock-42       [-1, 64, 8, 8]        0
          Conv2d-43       [-1, 64, 8, 8]     36,864
       BatchNorm2d-44       [-1, 64, 8, 8]       128
           ReLU-45       [-1, 64, 8, 8]        0
          Conv2d-46       [-1, 64, 8, 8]     36,864
       BatchNorm2d-47       [-1, 64, 8, 8]       128
           ReLU-48       [-1, 64, 8, 8]        0
      ResidualBlock-49       [-1, 64, 8, 8]        0
        AvgPool2d-50       [-1, 64, 1, 1]        0
          Linear-51          [-1, 10]       650
    ================================================================
    Total params: 195,738
    Trainable params: 195,738
    Non-trainable params: 0
    ----------------------------------------------------------------
    Input size (MB): 0.01
    Forward/backward pass size (MB): 3.63
    Params size (MB): 0.75
    Estimated Total Size (MB): 4.38
    ----------------------------------------------------------------

keras model summary

__________________________________________________________________________________________________
    Layer (type)          Output Shape     Param #   Connected to           
    ==================================================================================================
    input_26 (InputLayer)      (None, 32, 32, 3)  0                      
    __________________________________________________________________________________________________
    conv2d_103 (Conv2D)       (None, 32, 32, 16)  448     input_26[0][0]          
    __________________________________________________________________________________________________
    batch_normalization_99 (BatchNo (None, 32, 32, 16)  64     conv2d_103[0][0]         
    __________________________________________________________________________________________________
    activation_87 (Activation)   (None, 32, 32, 16)  0      batch_normalization_99[0][0]   
    __________________________________________________________________________________________________
    conv2d_104 (Conv2D)       (None, 32, 32, 16)  2320    activation_87[0][0]       
    __________________________________________________________________________________________________
    batch_normalization_100 (BatchN (None, 32, 32, 16)  64     conv2d_104[0][0]         
    __________________________________________________________________________________________________
    activation_88 (Activation)   (None, 32, 32, 16)  0      batch_normalization_100[0][0]  
    __________________________________________________________________________________________________
    conv2d_105 (Conv2D)       (None, 32, 32, 16)  2320    activation_88[0][0]       
    __________________________________________________________________________________________________
    batch_normalization_101 (BatchN (None, 32, 32, 16)  64     conv2d_105[0][0]         
    __________________________________________________________________________________________________
    add_34 (Add)          (None, 32, 32, 16)  0      activation_87[0][0]       
                                     batch_normalization_101[0][0]  
    __________________________________________________________________________________________________
    activation_89 (Activation)   (None, 32, 32, 16)  0      add_34[0][0]           
    __________________________________________________________________________________________________
    conv2d_106 (Conv2D)       (None, 32, 32, 16)  2320    activation_89[0][0]       
    __________________________________________________________________________________________________
    batch_normalization_102 (BatchN (None, 32, 32, 16)  64     conv2d_106[0][0]         
    __________________________________________________________________________________________________
    activation_90 (Activation)   (None, 32, 32, 16)  0      batch_normalization_102[0][0]  
    __________________________________________________________________________________________________
    conv2d_107 (Conv2D)       (None, 32, 32, 16)  2320    activation_90[0][0]       
    __________________________________________________________________________________________________
    batch_normalization_103 (BatchN (None, 32, 32, 16)  64     conv2d_107[0][0]         
    __________________________________________________________________________________________________
    add_35 (Add)          (None, 32, 32, 16)  0      activation_89[0][0]       
                                     batch_normalization_103[0][0]  
    __________________________________________________________________________________________________
    activation_91 (Activation)   (None, 32, 32, 16)  0      add_35[0][0]           
    __________________________________________________________________________________________________
    conv2d_108 (Conv2D)       (None, 16, 16, 32)  4640    activation_91[0][0]       
    __________________________________________________________________________________________________
    batch_normalization_104 (BatchN (None, 16, 16, 32)  128     conv2d_108[0][0]         
    __________________________________________________________________________________________________
    activation_92 (Activation)   (None, 16, 16, 32)  0      batch_normalization_104[0][0]  
    __________________________________________________________________________________________________
    conv2d_110 (Conv2D)       (None, 16, 16, 32)  4640    activation_91[0][0]       
    __________________________________________________________________________________________________
    conv2d_109 (Conv2D)       (None, 16, 16, 32)  9248    activation_92[0][0]       
    __________________________________________________________________________________________________
    batch_normalization_106 (BatchN (None, 16, 16, 32)  128     conv2d_110[0][0]         
    __________________________________________________________________________________________________
    batch_normalization_105 (BatchN (None, 16, 16, 32)  128     conv2d_109[0][0]         
    __________________________________________________________________________________________________
    add_36 (Add)          (None, 16, 16, 32)  0      batch_normalization_106[0][0]  
                                     batch_normalization_105[0][0]  
    __________________________________________________________________________________________________
    activation_93 (Activation)   (None, 16, 16, 32)  0      add_36[0][0]           
    __________________________________________________________________________________________________
    conv2d_111 (Conv2D)       (None, 16, 16, 32)  9248    activation_93[0][0]       
    __________________________________________________________________________________________________
    batch_normalization_107 (BatchN (None, 16, 16, 32)  128     conv2d_111[0][0]         
    __________________________________________________________________________________________________
    activation_94 (Activation)   (None, 16, 16, 32)  0      batch_normalization_107[0][0]  
    __________________________________________________________________________________________________
    conv2d_112 (Conv2D)       (None, 16, 16, 32)  9248    activation_94[0][0]       
    __________________________________________________________________________________________________
    batch_normalization_108 (BatchN (None, 16, 16, 32)  128     conv2d_112[0][0]         
    __________________________________________________________________________________________________
    add_37 (Add)          (None, 16, 16, 32)  0      activation_93[0][0]       
                                     batch_normalization_108[0][0]  
    __________________________________________________________________________________________________
    activation_95 (Activation)   (None, 16, 16, 32)  0      add_37[0][0]           
    __________________________________________________________________________________________________
    conv2d_113 (Conv2D)       (None, 8, 8, 64)   18496    activation_95[0][0]       
    __________________________________________________________________________________________________
    batch_normalization_109 (BatchN (None, 8, 8, 64)   256     conv2d_113[0][0]         
    __________________________________________________________________________________________________
    activation_96 (Activation)   (None, 8, 8, 64)   0      batch_normalization_109[0][0]  
    __________________________________________________________________________________________________
    conv2d_115 (Conv2D)       (None, 8, 8, 64)   18496    activation_95[0][0]       
    __________________________________________________________________________________________________
    conv2d_114 (Conv2D)       (None, 8, 8, 64)   36928    activation_96[0][0]       
    __________________________________________________________________________________________________
    batch_normalization_111 (BatchN (None, 8, 8, 64)   256     conv2d_115[0][0]         
    __________________________________________________________________________________________________
    batch_normalization_110 (BatchN (None, 8, 8, 64)   256     conv2d_114[0][0]         
    __________________________________________________________________________________________________
    add_38 (Add)          (None, 8, 8, 64)   0      batch_normalization_111[0][0]  
                                     batch_normalization_110[0][0]  
    __________________________________________________________________________________________________
    activation_97 (Activation)   (None, 8, 8, 64)   0      add_38[0][0]           
    __________________________________________________________________________________________________
    conv2d_116 (Conv2D)       (None, 8, 8, 64)   36928    activation_97[0][0]       
    __________________________________________________________________________________________________
    batch_normalization_112 (BatchN (None, 8, 8, 64)   256     conv2d_116[0][0]         
    __________________________________________________________________________________________________
    activation_98 (Activation)   (None, 8, 8, 64)   0      batch_normalization_112[0][0]  
    __________________________________________________________________________________________________
    conv2d_117 (Conv2D)       (None, 8, 8, 64)   36928    activation_98[0][0]       
    __________________________________________________________________________________________________
    batch_normalization_113 (BatchN (None, 8, 8, 64)   256     conv2d_117[0][0]         
    __________________________________________________________________________________________________
    add_39 (Add)          (None, 8, 8, 64)   0      activation_97[0][0]       
                                     batch_normalization_113[0][0]  
    __________________________________________________________________________________________________
    activation_99 (Activation)   (None, 8, 8, 64)   0      add_39[0][0]           
    __________________________________________________________________________________________________
    average_pooling2d_2 (AveragePoo (None, 1, 1, 64)   0      activation_99[0][0]       
    __________________________________________________________________________________________________
    flatten_2 (Flatten)       (None, 64)      0      average_pooling2d_2[0][0]    
    __________________________________________________________________________________________________
    dense_2 (Dense)         (None, 10)      650     flatten_2[0][0]         
    ==================================================================================================
    Total params: 197,418
    Trainable params: 196,298
    Non-trainable params: 1,120
    __________________________________________________________________________________________________

以上这篇keras的三种模型实现与区别说明就是小编分享给大家的全部内容了，希望能给大家一个参考，也希望大家多多支持菜鸟教程www.piaodoo.com。

[Link 1]: https://www.piaodoo.com/
[Link 2]: https://www.hnthzk.com/
[Link 3]: http://www.op-kg.com/