CNN模型复现5 ResNet50

最新推荐文章于 2024-07-04 08:58:47 发布
最新推荐文章于 2024-07-04 08:58:47 发布
阅读量411 收藏 1
点赞数
分类专栏： 计算机视觉
版权声明：本文为博主原创文章，遵循 CC 4.0 BY-SA 版权协议，转载请附上原文出处链接和本声明。
本文链接：https://blog.csdn.net/qq_35732321/article/details/115166725
版权
本文详细介绍了ResNet50网络结构，包括卷积、残差块、身份块和投影块的构建，并给出了使用Keras实现的完整代码。ResNet50通过残差学习解决深度网络中的梯度消失问题，通过平均池化而非全连接层获取最终特征，适用于图像分类等任务。
摘要由CSDN通过智能技术生成

目录

5.1 ResNet50

ResNet:arXiv: 1512.03385v1 [cs.CV] 10 Dec 2015 ;
链接: 论文下载link.
ResNet的设计有如下特点：

  1. 卷积+shortcut路径，构成一个Residual Block；
  2. ResNet中，通过conv的stride=2进行降采样；
  3. 通过Average Pooling得到最终的特征，而不是通过全连接层；
  4. 每个卷积层之后都紧接着BatchNorm layer。

(1) 网络结构
6. Conv + BatchNorm + Relu
7. ResNet blocks
8. AvgPool + Fully Connected layers
Mobile blocks

(2) Identity blocks
Identity block

  input 
    ↓
Conv 1x1 filter s=1
    BN  
   RelU
    ↓
Conv 3x3 filter s=1
    BN
   RelU
    ↓
Conv 1x1 filter*4 s=1
    BN
    ↓
    + input
   RelU

(3) Projection blocks

  input 
    ↓
Conv 1x1, filter, s=1
    BN
   RelU
    ↓
Conv 3x3 ,filter1 ,s=2
    BN
   RelU
    ↓
Conv 1x1, filter*4, s=1
    BN
    ↓
    + Conv(k=1,filter*4, s=2)(input)
   RelU

(4) Model diagram
RestNet50

Conv(f=32,k=3,s=2,p=s)
BatchNormalization
ReLU
MaxPool2D
    ↓
resNet_block(x,f=64,r=3,s=1)
resNet_block(x,f=128,r=4,s=2)
resNet_block(x,f=256,r=6,s=2)
resNet_block(x,f=512,r=6,s=2)
    ↓
GlobalAvgPool2D()(x)
Dense(1000)
softmax

5.2 代码

(1) 代码流程

  1. 导入程序包
  2. 编写Conv + BatchNorm + Relu block
  3. 建立Identity blocks
  4. 建立Projection blocks
  5. 建立ResNet blocks
  6. 建立model

(2) 代码

# 1 Import
from keras import Model
from keras.utils import plot_model
from keras.layers import Input,Conv2D,BatchNormalization,ReLU,Add,MaxPool2D,GlobalAvgPool2D,Dense


# 2 Conv-BatchNorm-ReLU blocks
def conv_bachnorm_relu(x,filters,kernel_size,strides):
    x = Conv2D(filters=filters,kernel_size=kernel_size,strides=strides,padding='same')(x)
    x = BatchNormalization()(x)
    x = ReLU()(x)
    return x



# 3 Identity blocks
def identity_block(input,filters):
    x = conv_bachnorm_relu(input,filters,kernel_size=1,strides=1)
    x = conv_bachnorm_relu(x,filters,kernel_size=3,strides=1)
    x = Conv2D(filters=filters*4,kernel_size=1,strides=1)(x)
    x = BatchNormalization()(x)
    x = Add()([x,input])
    x = ReLU()(x)
    return x

# 4 Projection blocks
def projection_block(input,filters,strides):
    x = conv_bachnorm_relu(input,filters=filters,kernel_size=1,strides=1)
    x = conv_bachnorm_relu(x,filters=filters,kernel_size=3,strides=strides)
    x = Conv2D(filters=filters*4,kernel_size=1,strides=1)(x)
    x = BatchNormalization()(x)

    shortcut = Conv2D(filters=filters*4,kernel_size=1,strides=strides)(input)
    shortcut = BatchNormalization()(shortcut)
    x = Add()([x,shortcut])
    x = ReLU()(x)
    return x


# 5 ResNet blocks
def resnet_block(x,filters,reps,strides):
    x = projection_block(x,filters=filters,strides=strides)
    for _ in range(reps-1):
        x = identity_block(x,filters=filters)
    return x


# 6 Model
input = Input((224,224,3))
x = conv_bachnorm_relu(input,filters=64,kernel_size=7,strides=2) # (None, 224,224,3)-->(None, 112, 112, 64)
x = MaxPool2D(pool_size=3,strides=2,padding='same')(x)  # (None, 112, 112, 64)-->(None, 56, 56, 64)
x = resnet_block(x,filters=64,reps=3,strides=1)   # (None, 56, 56, 64) --> (None, 56, 56, 256)
x = resnet_block(x,filters=128,reps=4,strides=2)  # (None, 56, 56, 256) --> (None,28, 28, 512)
x = resnet_block(x,filters=256,reps=6,strides=2)  # (None, 28, 28, 512) --> (None, 14, 14, 1024)
x = resnet_block(x,filters=512,reps=3,strides=2)  # (None, 14, 14, 1024) -->  (None, 7, 7, 2048)
x = GlobalAvgPool2D()(x)       # (None, 2048)
output = Dense(units=1000,activation='softmax')(x) # (None, 1000)

model = Model(inputs=input,outputs=output)
model.summary()

'''
__________________________________________________________________________________________________
Layer (type)                    Output Shape         Param #     Connected to                     
==================================================================================================
input_1 (InputLayer)            (None, 224, 224, 3)  0                                            
__________________________________________________________________________________________________
conv2d_1 (Conv2D)               (None, 112, 112, 64) 9472        input_1[0][0]                    
__________________________________________________________________________________________________
batch_normalization_1 (BatchNor (None, 112, 112, 64) 256         conv2d_1[0][0]                   
__________________________________________________________________________________________________
re_lu_1 (ReLU)                  (None, 112, 112, 64) 0           batch_normalization_1[0][0]      
__________________________________________________________________________________________________
max_pooling2d_1 (MaxPooling2D)  (None, 56, 56, 64)   0           re_lu_1[0][0]                    
__________________________________________________________________________________________________
conv2d_2 (Conv2D)               (None, 56, 56, 64)   4160        max_pooling2d_1[0][0]            
__________________________________________________________________________________________________
batch_normalization_2 (BatchNor (None, 56, 56, 64)   256         conv2d_2[0][0]                   
__________________________________________________________________________________________________
re_lu_2 (ReLU)                  (None, 56, 56, 64)   0           batch_normalization_2[0][0]      
__________________________________________________________________________________________________
conv2d_3 (Conv2D)               (None, 56, 56, 64)   36928       re_lu_2[0][0]                    
__________________________________________________________________________________________________
batch_normalization_3 (BatchNor (None, 56, 56, 64)   256         conv2d_3[0][0]                   
__________________________________________________________________________________________________
re_lu_3 (ReLU)                  (None, 56, 56, 64)   0           batch_normalization_3[0][0]      
__________________________________________________________________________________________________
conv2d_4 (Conv2D)               (None, 56, 56, 256)  16640       re_lu_3[0][0]                    
__________________________________________________________________________________________________
conv2d_5 (Conv2D)               (None, 56, 56, 256)  16640       max_pooling2d_1[0][0]            
__________________________________________________________________________________________________
batch_normalization_4 (BatchNor (None, 56, 56, 256)  1024        conv2d_4[0][0]                   
__________________________________________________________________________________________________
batch_normalization_5 (BatchNor (None, 56, 56, 256)  1024        conv2d_5[0][0]                   
__________________________________________________________________________________________________
add_1 (Add)                     (None, 56, 56, 256)  0           batch_normalization_4[0][0]      
                                                                 batch_normalization_5[0][0]      
__________________________________________________________________________________________________
re_lu_4 (ReLU)                  (None, 56, 56, 256)  0           add_1[0][0]                      
__________________________________________________________________________________________________
conv2d_6 (Conv2D)               (None, 56, 56, 64)   16448       re_lu_4[0][0]                    
__________________________________________________________________________________________________
batch_normalization_6 (BatchNor (None, 56, 56, 64)   256         conv2d_6[0][0]                   
__________________________________________________________________________________________________
re_lu_5 (ReLU)                  (None, 56, 56, 64)   0           batch_normalization_6[0][0]      
__________________________________________________________________________________________________
conv2d_7 (Conv2D)               (None, 56, 56, 64)   36928       re_lu_5[0][0]                    
__________________________________________________________________________________________________
batch_normalization_7 (BatchNor (None, 56, 56, 64)   256         conv2d_7[0][0]                   
__________________________________________________________________________________________________
re_lu_6 (ReLU)                  (None, 56, 56, 64)   0           batch_normalization_7[0][0]      
__________________________________________________________________________________________________
conv2d_8 (Conv2D)               (None, 56, 56, 256)  16640       re_lu_6[0][0]                    
__________________________________________________________________________________________________
batch_normalization_8 (BatchNor (None, 56, 56, 256)  1024        conv2d_8[0][0]                   
__________________________________________________________________________________________________
add_2 (Add)                     (None, 56, 56, 256)  0           batch_normalization_8[0][0]      
                                                                 re_lu_4[0][0]                    
__________________________________________________________________________________________________
re_lu_7 (ReLU)                  (None, 56, 56, 256)  0           add_2[0][0]                      
__________________________________________________________________________________________________
conv2d_9 (Conv2D)               (None, 56, 56, 64)   16448       re_lu_7[0][0]                    
__________________________________________________________________________________________________
batch_normalization_9 (BatchNor (None, 56, 56, 64)   256         conv2d_9[0][0]                   
__________________________________________________________________________________________________
re_lu_8 (ReLU)                  (None, 56, 56, 64)   0           batch_normalization_9[0][0]      
__________________________________________________________________________________________________
conv2d_10 (Conv2D)              (None, 56, 56, 64)   36928       re_lu_8[0][0]                    
__________________________________________________________________________________________________
batch_normalization_10 (BatchNo (None, 56, 56, 64)   256         conv2d_10[0][0]                  
__________________________________________________________________________________________________
re_lu_9 (ReLU)                  (None, 56, 56, 64)   0           batch_normalization_10[0][0]     
__________________________________________________________________________________________________
conv2d_11 (Conv2D)              (None, 56, 56, 256)  16640       re_lu_9[0][0]                    
__________________________________________________________________________________________________
batch_normalization_11 (BatchNo (None, 56, 56, 256)  1024        conv2d_11[0][0]                  
__________________________________________________________________________________________________
add_3 (Add)                     (None, 56, 56, 256)  0           batch_normalization_11[0][0]     
                                                                 re_lu_7[0][0]                    
__________________________________________________________________________________________________
re_lu_10 (ReLU)                 (None, 56, 56, 256)  0           add_3[0][0]                      
__________________________________________________________________________________________________
conv2d_12 (Conv2D)              (None, 56, 56, 128)  32896       re_lu_10[0][0]                   
__________________________________________________________________________________________________
batch_normalization_12 (BatchNo (None, 56, 56, 128)  512         conv2d_12[0][0]                  
__________________________________________________________________________________________________
re_lu_11 (ReLU)                 (None, 56, 56, 128)  0           batch_normalization_12[0][0]     
__________________________________________________________________________________________________
conv2d_13 (Conv2D)              (None, 28, 28, 128)  147584      re_lu_11[0][0]                   
__________________________________________________________________________________________________
batch_normalization_13 (BatchNo (None, 28, 28, 128)  512         conv2d_13[0][0]                  
__________________________________________________________________________________________________
re_lu_12 (ReLU)                 (None, 28, 28, 128)  0           batch_normalization_13[0][0]     
__________________________________________________________________________________________________
conv2d_14 (Conv2D)              (None, 28, 28, 512)  66048       re_lu_12[0][0]                   
__________________________________________________________________________________________________
conv2d_15 (Conv2D)              (None, 28, 28, 512)  131584      re_lu_10[0][0]                   
__________________________________________________________________________________________________
batch_normalization_14 (BatchNo (None, 28, 28, 512)  2048        conv2d_14[0][0]                  
__________________________________________________________________________________________________
batch_normalization_15 (BatchNo (None, 28, 28, 512)  2048        conv2d_15[0][0]                  
__________________________________________________________________________________________________
add_4 (Add)                     (None, 28, 28, 512)  0           batch_normalization_14[0][0]     
                                                                 batch_normalization_15[0][0]     
__________________________________________________________________________________________________
re_lu_13 (ReLU)                 (None, 28, 28, 512)  0           add_4[0][0]                      
__________________________________________________________________________________________________
conv2d_16 (Conv2D)              (None, 28, 28, 128)  65664       re_lu_13[0][0]                   
__________________________________________________________________________________________________
batch_normalization_16 (BatchNo (None, 28, 28, 128)  512         conv2d_16[0][0]                  
__________________________________________________________________________________________________
re_lu_14 (ReLU)                 (None, 28, 28, 128)  0           batch_normalization_16[0][0]     
__________________________________________________________________________________________________
conv2d_17 (Conv2D)              (None, 28, 28, 128)  147584      re_lu_14[0][0]                   
__________________________________________________________________________________________________
batch_normalization_17 (BatchNo (None, 28, 28, 128)  512         conv2d_17[0][0]                  
__________________________________________________________________________________________________
re_lu_15 (ReLU)                 (None, 28, 28, 128)  0           batch_normalization_17[0][0]     
__________________________________________________________________________________________________
conv2d_18 (Conv2D)              (None, 28, 28, 512)  66048       re_lu_15[0][0]                   
__________________________________________________________________________________________________
batch_normalization_18 (BatchNo (None, 28, 28, 512)  2048        conv2d_18[0][0]                  
__________________________________________________________________________________________________
add_5 (Add)                     (None, 28, 28, 512)  0           batch_normalization_18[0][0]     
                                                                 re_lu_13[0][0]                   
__________________________________________________________________________________________________
re_lu_16 (ReLU)                 (None, 28, 28, 512)  0           add_5[0][0]                      
__________________________________________________________________________________________________
conv2d_19 (Conv2D)              (None, 28, 28, 128)  65664       re_lu_16[0][0]                   
__________________________________________________________________________________________________
batch_normalization_19 (BatchNo (None, 28, 28, 128)  512         conv2d_19[0][0]                  
__________________________________________________________________________________________________
re_lu_17 (ReLU)                 (None, 28, 28, 128)  0           batch_normalization_19[0][0]     
__________________________________________________________________________________________________
conv2d_20 (Conv2D)              (None, 28, 28, 128)  147584      re_lu_17[0][0]                   
__________________________________________________________________________________________________
batch_normalization_20 (BatchNo (None, 28, 28, 128)  512         conv2d_20[0][0]                  
__________________________________________________________________________________________________
re_lu_18 (ReLU)                 (None, 28, 28, 128)  0           batch_normalization_20[0][0]     
__________________________________________________________________________________________________
conv2d_21 (Conv2D)              (None, 28, 28, 512)  66048       re_lu_18[0][0]                   
__________________________________________________________________________________________________
batch_normalization_21 (BatchNo (None, 28, 28, 512)  2048        conv2d_21[0][0]                  
__________________________________________________________________________________________________
add_6 (Add)                     (None, 28, 28, 512)  0           batch_normalization_21[0][0]     
                                                                 re_lu_16[0][0]                   
__________________________________________________________________________________________________
re_lu_19 (ReLU)                 (None, 28, 28, 512)  0           add_6[0][0]                      
__________________________________________________________________________________________________
conv2d_22 (Conv2D)              (None, 28, 28, 128)  65664       re_lu_19[0][0]                   
__________________________________________________________________________________________________
batch_normalization_22 (BatchNo (None, 28, 28, 128)  512         conv2d_22[0][0]                  
__________________________________________________________________________________________________
re_lu_20 (ReLU)                 (None, 28, 28, 128)  0           batch_normalization_22[0][0]     
__________________________________________________________________________________________________
conv2d_23 (Conv2D)              (None, 28, 28, 128)  147584      re_lu_20[0][0]                   
__________________________________________________________________________________________________
batch_normalization_23 (BatchNo (None, 28, 28, 128)  512         conv2d_23[0][0]                  
__________________________________________________________________________________________________
re_lu_21 (ReLU)                 (None, 28, 28, 128)  0           batch_normalization_23[0][0]     
__________________________________________________________________________________________________
conv2d_24 (Conv2D)              (None, 28, 28, 512)  66048       re_lu_21[0][0]                   
__________________________________________________________________________________________________
batch_normalization_24 (BatchNo (None, 28, 28, 512)  2048        conv2d_24[0][0]                  
__________________________________________________________________________________________________
add_7 (Add)                     (None, 28, 28, 512)  0           batch_normalization_24[0][0]     
                                                                 re_lu_19[0][0]                   
__________________________________________________________________________________________________
re_lu_22 (ReLU)                 (None, 28, 28, 512)  0           add_7[0][0]                      
__________________________________________________________________________________________________
conv2d_25 (Conv2D)              (None, 28, 28, 256)  131328      re_lu_22[0][0]                   
__________________________________________________________________________________________________
batch_normalization_25 (BatchNo (None, 28, 28, 256)  1024        conv2d_25[0][0]                  
__________________________________________________________________________________________________
re_lu_23 (ReLU)                 (None, 28, 28, 256)  0           batch_normalization_25[0][0]     
__________________________________________________________________________________________________
conv2d_26 (Conv2D)              (None, 14, 14, 256)  590080      re_lu_23[0][0]                   
__________________________________________________________________________________________________
batch_normalization_26 (BatchNo (None, 14, 14, 256)  1024        conv2d_26[0][0]                  
__________________________________________________________________________________________________
re_lu_24 (ReLU)                 (None, 14, 14, 256)  0           batch_normalization_26[0][0]     
__________________________________________________________________________________________________
conv2d_27 (Conv2D)              (None, 14, 14, 1024) 263168      re_lu_24[0][0]                   
__________________________________________________________________________________________________
conv2d_28 (Conv2D)              (None, 14, 14, 1024) 525312      re_lu_22[0][0]                   
__________________________________________________________________________________________________
batch_normalization_27 (BatchNo (None, 14, 14, 1024) 4096        conv2d_27[0][0]                  
__________________________________________________________________________________________________
batch_normalization_28 (BatchNo (None, 14, 14, 1024) 4096        conv2d_28[0][0]                  
__________________________________________________________________________________________________
add_8 (Add)                     (None, 14, 14, 1024) 0           batch_normalization_27[0][0]     
                                                                 batch_normalization_28[0][0]     
__________________________________________________________________________________________________
re_lu_25 (ReLU)                 (None, 14, 14, 1024) 0           add_8[0][0]                      
__________________________________________________________________________________________________
conv2d_29 (Conv2D)              (None, 14, 14, 256)  262400      re_lu_25[0][0]                   
__________________________________________________________________________________________________
batch_normalization_29 (BatchNo (None, 14, 14, 256)  1024        conv2d_29[0][0]                  
__________________________________________________________________________________________________
re_lu_26 (ReLU)                 (None, 14, 14, 256)  0           batch_normalization_29[0][0]     
__________________________________________________________________________________________________
conv2d_30 (Conv2D)              (None, 14, 14, 256)  590080      re_lu_26[0][0]                   
__________________________________________________________________________________________________
batch_normalization_30 (BatchNo (None, 14, 14, 256)  1024        conv2d_30[0][0]                  
__________________________________________________________________________________________________
re_lu_27 (ReLU)                 (None, 14, 14, 256)  0           batch_normalization_30[0][0]     
__________________________________________________________________________________________________
conv2d_31 (Conv2D)              (None, 14, 14, 1024) 263168      re_lu_27[0][0]                   
__________________________________________________________________________________________________
batch_normalization_31 (BatchNo (None, 14, 14, 1024) 4096        conv2d_31[0][0]                  
__________________________________________________________________________________________________
add_9 (Add)                     (None, 14, 14, 1024) 0           batch_normalization_31[0][0]     
                                                                 re_lu_25[0][0]                   
__________________________________________________________________________________________________
re_lu_28 (ReLU)                 (None, 14, 14, 1024) 0           add_9[0][0]                      
__________________________________________________________________________________________________
conv2d_32 (Conv2D)              (None, 14, 14, 256)  262400      re_lu_28[0][0]                   
__________________________________________________________________________________________________
batch_normalization_32 (BatchNo (None, 14, 14, 256)  1024        conv2d_32[0][0]                  
__________________________________________________________________________________________________
re_lu_29 (ReLU)                 (None, 14, 14, 256)  0           batch_normalization_32[0][0]     
__________________________________________________________________________________________________
conv2d_33 (Conv2D)              (None, 14, 14, 256)  590080      re_lu_29[0][0]                   
__________________________________________________________________________________________________
batch_normalization_33 (BatchNo (None, 14, 14, 256)  1024        conv2d_33[0][0]                  
__________________________________________________________________________________________________
re_lu_30 (ReLU)                 (None, 14, 14, 256)  0           batch_normalization_33[0][0]     
__________________________________________________________________________________________________
conv2d_34 (Conv2D)              (None, 14, 14, 1024) 263168      re_lu_30[0][0]                   
__________________________________________________________________________________________________
batch_normalization_34 (BatchNo (None, 14, 14, 1024) 4096        conv2d_34[0][0]                  
__________________________________________________________________________________________________
add_10 (Add)                    (None, 14, 14, 1024) 0           batch_normalization_34[0][0]     
                                                                 re_lu_28[0][0]                   
__________________________________________________________________________________________________
re_lu_31 (ReLU)                 (None, 14, 14, 1024) 0           add_10[0][0]                     
__________________________________________________________________________________________________
conv2d_35 (Conv2D)              (None, 14, 14, 256)  262400      re_lu_31[0][0]                   
__________________________________________________________________________________________________
batch_normalization_35 (BatchNo (None, 14, 14, 256)  1024        conv2d_35[0][0]                  
__________________________________________________________________________________________________
re_lu_32 (ReLU)                 (None, 14, 14, 256)  0           batch_normalization_35[0][0]     
__________________________________________________________________________________________________
conv2d_36 (Conv2D)              (None, 14, 14, 256)  590080      re_lu_32[0][0]                   
__________________________________________________________________________________________________
batch_normalization_36 (BatchNo (None, 14, 14, 256)  1024        conv2d_36[0][0]                  
__________________________________________________________________________________________________
re_lu_33 (ReLU)                 (None, 14, 14, 256)  0           batch_normalization_36[0][0]     
__________________________________________________________________________________________________
conv2d_37 (Conv2D)              (None, 14, 14, 1024) 263168      re_lu_33[0][0]                   
__________________________________________________________________________________________________
batch_normalization_37 (BatchNo (None, 14, 14, 1024) 4096        conv2d_37[0][0]                  
__________________________________________________________________________________________________
add_11 (Add)                    (None, 14, 14, 1024) 0           batch_normalization_37[0][0]     
                                                                 re_lu_31[0][0]                   
__________________________________________________________________________________________________
re_lu_34 (ReLU)                 (None, 14, 14, 1024) 0           add_11[0][0]                     
__________________________________________________________________________________________________
conv2d_38 (Conv2D)              (None, 14, 14, 256)  262400      re_lu_34[0][0]                   
__________________________________________________________________________________________________
batch_normalization_38 (BatchNo (None, 14, 14, 256)  1024        conv2d_38[0][0]                  
__________________________________________________________________________________________________
re_lu_35 (ReLU)                 (None, 14, 14, 256)  0           batch_normalization_38[0][0]     
__________________________________________________________________________________________________
conv2d_39 (Conv2D)              (None, 14, 14, 256)  590080      re_lu_35[0][0]                   
__________________________________________________________________________________________________
batch_normalization_39 (BatchNo (None, 14, 14, 256)  1024        conv2d_39[0][0]                  
__________________________________________________________________________________________________
re_lu_36 (ReLU)                 (None, 14, 14, 256)  0           batch_normalization_39[0][0]     
__________________________________________________________________________________________________
conv2d_40 (Conv2D)              (None, 14, 14, 1024) 263168      re_lu_36[0][0]                   
__________________________________________________________________________________________________
batch_normalization_40 (BatchNo (None, 14, 14, 1024) 4096        conv2d_40[0][0]                  
__________________________________________________________________________________________________
add_12 (Add)                    (None, 14, 14, 1024) 0           batch_normalization_40[0][0]     
                                                                 re_lu_34[0][0]                   
__________________________________________________________________________________________________
re_lu_37 (ReLU)                 (None, 14, 14, 1024) 0           add_12[0][0]                     
__________________________________________________________________________________________________
conv2d_41 (Conv2D)              (None, 14, 14, 256)  262400      re_lu_37[0][0]                   
__________________________________________________________________________________________________
batch_normalization_41 (BatchNo (None, 14, 14, 256)  1024        conv2d_41[0][0]                  
__________________________________________________________________________________________________
re_lu_38 (ReLU)                 (None, 14, 14, 256)  0           batch_normalization_41[0][0]     
__________________________________________________________________________________________________
conv2d_42 (Conv2D)              (None, 14, 14, 256)  590080      re_lu_38[0][0]                   
__________________________________________________________________________________________________
batch_normalization_42 (BatchNo (None, 14, 14, 256)  1024        conv2d_42[0][0]                  
__________________________________________________________________________________________________
re_lu_39 (ReLU)                 (None, 14, 14, 256)  0           batch_normalization_42[0][0]     
__________________________________________________________________________________________________
conv2d_43 (Conv2D)              (None, 14, 14, 1024) 263168      re_lu_39[0][0]                   
__________________________________________________________________________________________________
batch_normalization_43 (BatchNo (None, 14, 14, 1024) 4096        conv2d_43[0][0]                  
__________________________________________________________________________________________________
add_13 (Add)                    (None, 14, 14, 1024) 0           batch_normalization_43[0][0]     
                                                                 re_lu_37[0][0]                   
__________________________________________________________________________________________________
re_lu_40 (ReLU)                 (None, 14, 14, 1024) 0           add_13[0][0]                     
__________________________________________________________________________________________________
conv2d_44 (Conv2D)              (None, 14, 14, 512)  524800      re_lu_40[0][0]                   
__________________________________________________________________________________________________
batch_normalization_44 (BatchNo (None, 14, 14, 512)  2048        conv2d_44[0][0]                  
__________________________________________________________________________________________________
re_lu_41 (ReLU)                 (None, 14, 14, 512)  0           batch_normalization_44[0][0]     
__________________________________________________________________________________________________
conv2d_45 (Conv2D)              (None, 7, 7, 512)    2359808     re_lu_41[0][0]                   
__________________________________________________________________________________________________
batch_normalization_45 (BatchNo (None, 7, 7, 512)    2048        conv2d_45[0][0]                  
__________________________________________________________________________________________________
re_lu_42 (ReLU)                 (None, 7, 7, 512)    0           batch_normalization_45[0][0]     
__________________________________________________________________________________________________
conv2d_46 (Conv2D)              (None, 7, 7, 2048)   1050624     re_lu_42[0][0]                   
__________________________________________________________________________________________________
conv2d_47 (Conv2D)              (None, 7, 7, 2048)   2099200     re_lu_40[0][0]                   
__________________________________________________________________________________________________
batch_normalization_46 (BatchNo (None, 7, 7, 2048)   8192        conv2d_46[0][0]                  
__________________________________________________________________________________________________
batch_normalization_47 (BatchNo (None, 7, 7, 2048)   8192        conv2d_47[0][0]                  
__________________________________________________________________________________________________
add_14 (Add)                    (None, 7, 7, 2048)   0           batch_normalization_46[0][0]     
                                                                 batch_normalization_47[0][0]     
__________________________________________________________________________________________________
re_lu_43 (ReLU)                 (None, 7, 7, 2048)   0           add_14[0][0]                     
__________________________________________________________________________________________________
conv2d_48 (Conv2D)              (None, 7, 7, 512)    1049088     re_lu_43[0][0]                   
__________________________________________________________________________________________________
batch_normalization_48 (BatchNo (None, 7, 7, 512)    2048        conv2d_48[0][0]                  
__________________________________________________________________________________________________
re_lu_44 (ReLU)                 (None, 7, 7, 512)    0           batch_normalization_48[0][0]     
__________________________________________________________________________________________________
conv2d_49 (Conv2D)              (None, 7, 7, 512)    2359808     re_lu_44[0][0]                   
__________________________________________________________________________________________________
batch_normalization_49 (BatchNo (None, 7, 7, 512)    2048        conv2d_49[0][0]                  
__________________________________________________________________________________________________
re_lu_45 (ReLU)                 (None, 7, 7, 512)    0           batch_normalization_49[0][0]     
__________________________________________________________________________________________________
conv2d_50 (Conv2D)              (None, 7, 7, 2048)   1050624     re_lu_45[0][0]                   
__________________________________________________________________________________________________
batch_normalization_50 (BatchNo (None, 7, 7, 2048)   8192        conv2d_50[0][0]                  
__________________________________________________________________________________________________
add_15 (Add)                    (None, 7, 7, 2048)   0           batch_normalization_50[0][0]     
                                                                 re_lu_43[0][0]                   
__________________________________________________________________________________________________
re_lu_46 (ReLU)                 (None, 7, 7, 2048)   0           add_15[0][0]                     
__________________________________________________________________________________________________
conv2d_51 (Conv2D)              (None, 7, 7, 512)    1049088     re_lu_46[0][0]                   
__________________________________________________________________________________________________
batch_normalization_51 (BatchNo (None, 7, 7, 512)    2048        conv2d_51[0][0]                  
__________________________________________________________________________________________________
re_lu_47 (ReLU)                 (None, 7, 7, 512)    0           batch_normalization_51[0][0]     
__________________________________________________________________________________________________
conv2d_52 (Conv2D)              (None, 7, 7, 512)    2359808     re_lu_47[0][0]                   
__________________________________________________________________________________________________
batch_normalization_52 (BatchNo (None, 7, 7, 512)    2048        conv2d_52[0][0]                  
__________________________________________________________________________________________________
re_lu_48 (ReLU)                 (None, 7, 7, 512)    0           batch_normalization_52[0][0]     
__________________________________________________________________________________________________
conv2d_53 (Conv2D)              (None, 7, 7, 2048)   1050624     re_lu_48[0][0]                   
__________________________________________________________________________________________________
batch_normalization_53 (BatchNo (None, 7, 7, 2048)   8192        conv2d_53[0][0]                  
__________________________________________________________________________________________________
add_16 (Add)                    (None, 7, 7, 2048)   0           batch_normalization_53[0][0]     
                                                                 re_lu_46[0][0]                   
__________________________________________________________________________________________________
re_lu_49 (ReLU)                 (None, 7, 7, 2048)   0           add_16[0][0]                     
__________________________________________________________________________________________________
global_average_pooling2d_1 (Glo (None, 2048)         0           re_lu_49[0][0]                   
__________________________________________________________________________________________________
dense_1 (Dense)                 (None, 1000)         2049000     global_average_pooling2d_1[0][0] 
==================================================================================================
Total params: 25,636,712
Trainable params: 25,583,592
Non-trainable params: 53,120
__________________________________________________________________________________________________

Process finished with exit code 0
'''
山居秋暝LS
关注
专栏目录
比较TensorFlow中全连接，简单CNN和ResNet50在二进制图像分类中的性能
weixin_26632369的博客
08-25 1045
In the world of machine learning, there are three models that you can use to perform binary image classification: a fully-connected network, a convolutional neural network, or a pre-trained network li...
你必须知道的 | ResNet神经网络模型复现
m0_60673947的博客
10-12 478
一、模型介绍 ResNet由Deep Residual Learning for Image Recognition提出，现在在深度学习网络中应用越来越广泛。主要因为模型训练的时候产生梯度消失/爆炸，导致模型难以收敛，尽管batch normalization的使用可以使几十层的网络使用SGD算法训练收敛，但是更深的网络就没办法了。就算能收敛，也还是存在模型性能退化的问题，模型误差不降反升。为了解决这个问题，论文作者提出残差结构，让网络不去直接拟合目标函数H(x)，而是去拟合它的残差F(x)=H(x)+x
Faster RCNN ResNet-50
12-26
适用于caffe框架下py-faster-rcnn的resnet-50的prototxt.
经典CNN(一）：ResNet-50算法实战与解析
ali1174的专栏
07-13 9256
残差网络是为了解决神经网络隐藏层过多时，而引起的网络退化问题。退化（degradation）问题是指：当网络隐藏层变多时，网络的准确度达到饱和，然后急剧退化，而且这个退化不是由于过拟合引起的。拓展：深度神经网络的“两朵乌云”梯度弥散/爆炸简单来讲就是网络太深了，会导致模型训练难以收敛。这个问题可以被标准初始化和中间层正规化的方法有效控制。网络退化随着网络深度增加，网络的表现先是逐渐增加至饱和，然后迅速下降，这个退化不是由于过拟合而引起的。
【深度学习】ResNet——CNN经典网络模型详解(pytorch实现)
fengdu78的博客
09-02 3812
建议大家可以实践下，代码都很详细，有不清楚的地方评论区见~1、前言ResNet（Residual Neural Network）由微软研究院的Kaiming He等四名华人提出，通过使用...
CNN算法（一）——残差网络ResNet-50
最新发布
m0_58378112的博客
07-04 1985
根据任务选择合适的模型：例如图像分类可能使用CNN，序列任务可能使用RNN或Transformer。应用正则化技术：L1、L2正则化或Dropout来减少过拟合。使用批量归一化：来加速训练并提高模型稳定性。模型集成：训练多个模型并将其集成起来，以提高整体性能。多任务学习：如果可能，使用多任务学习同时训练模型完成多个相关任务。损失函数选择：根据具体任务选择合适的损失函数，如均方误差、交叉熵等。
TensorFlow 复现ResNet系列模型
qq_26063799的博客
10-10 3642
阅前须知： 为了使本文结构精简，理解简单，所以会尽量少涉及到有关数学公式，降低学习门槛，带领读者快速搭建ResNet-152经典模型并投入训练。 本文的最后会放出博主自己复现的ResNet模型，投入自己的数据集进行训练。如读者在阅读时发现有错误的地方欢在评论的地方指出，共同进步 编译环境：Python3.5                     TensorFlow-gpu 1.3.0 一、...
基于megengine的freeanchor的目标检测模型，主干网络ResNet50
10-17
在这个场景中，我们关注的是一个基于MegEngine框架实现的FreeAnchor目标检测模型，该模型的主干网络采用了经典的ResNet50架构。下面我们将深入探讨这些概念。 **MegEngine**: MegEngine 是一个高性能的深度学习框架...
Caffe下faster-rcnn的resnet50的配置文件
12-19
总而言之，Faster R-CNN结合ResNet-50在Caffe中的配置涉及到深度学习模型的构建、训练和测试策略，这些配置文件对于理解和复现目标检测实验至关重要。通过精细调整这些参数，研究者和开发者可以优化模型的性能，以...
经典CNN算法解析实战-第J2周：ResNet50V2算法实战与解析
Oaix的博客
02-16 646
第J2周：ResNet50V2算法实战与解析
resnet_v1_50模型
02-24
这是一个resnet_v1_50的模型，slim微调时可以对该模型的进行重新训练，同时减少了训练的时间
pytorch逐行搭建CNN系列（三）ResNet50
nszzzzdr的博客
03-22 1713
在这里插入图片描述](https://img-blog.csdnimg.cn/cda0a32ed5ec48d1a1b58a8377783483.png。提出Batch Normalization 加速训练（丢弃dropout）:将一批数据的feature map转化为满足均值=0，方差=1的分布。提出了残差网络块（Residual）：人为地让神经网络某些层跳过下一层神经元的连接，隔层相连，弱化每层之间的强联系。利用summary()函数打印网络结构和参数。使用了一种shortcut的跳跃连接方式。
深度学习J1周-ResNet50算法实战与解析_鸟类识别（CNN）
m0_62237233的博客
03-06 1473
残差网络是为了解决神经网络隐藏层过多时，而引起的网络退化问题。退化（degradation）问题是指：当网络隐藏层变多时，网络的准确度达到饱和然后急剧退化，而且这个退化不是由于过拟合引起的。拓展： 深度神经网络的“两朵乌云”●梯度弥散/爆炸简单来讲就是网络太深了，会导致模型训练难以收敛。这个问题可以被标准初始化和中间层正规化的方法有效控制。（现阶段知道这么一回事就好了）●网络退化随着网络深度增加，网络的表现先是逐渐增加至饱和，然后迅速下降，这个退化不是由于过拟合引起的。
CNN与ResNet联系与区别
geo436872的博客
06-03 7606
在传统的CNN中，信息从输入一直通过各个层向前传播，而ResNet则在传统CNN基础上增加了残差块，在ResNet中，每个残差块包含了多个卷积层和一条跨过若干层的“捷径”（shortcut connection），使得信息可以直接从前面的某一层传到后面的某一层。由于残差块的使用，ResNet可以训练更深的神经网络，例如ResNet-152就有152个卷积层，并且在训练误差和测试误差上都表现出了更好的性能。而ResNet是残差网络，是一种在CNN基础上的变体，通过引入残差块来训练更深层次的神经网络。
Resnet50
afadgas的博客
10-08 1748
resnet50
ResNet——CNN经典网络模型详解(pytorch实现)
热门推荐
weixin_44023658的博客
04-29 8万+
1、前言 ResNet（Residual Neural Network）由微软研究院的Kaiming He等四名华人提出，通过使用ResNet Unit成功训练出了152层的神经网络，并在ILSVRC2015比赛中取得冠军，在top5上的错误率为3.57%，同时参数量比VGGNet低，效果非常突出。ResNet的结构可以极快的加速神经网络的训练，模型的准确率也有比较大的提升。同时ResNet的推广...
resnet50 网络结构全图
AI与算法都要通俗易懂
06-26 3465
可视化CNN， AI算法的搬砖真实性！
迁移学习resnet_ResNet-V1-50卷积神经网络迁移学习进行不同品种的花的分类识别
weixin_32921457的博客
12-23 907
运行环境python3.6.3、tensorflow1.10.0Intel@AIDevCloud：Intel Xeon Gold 6128 processors集群数据和模型来源思路数据集分析及处理数据集文件解压后共有五个文件夹，每个文件夹都包含一定数量的花的图片，一个文件夹对应一个品种，图片各种尺寸都有，均为jpg格式，均为彩色图片。这里利用tensorflow提供的图片处理工具将所有图片转为3...
resnet_v1_50源码的理解与分析
Huang_Fj的博客
09-10 3345
源代码链接：https://github.com/tensorflow/models/blob/master/research/slim/nets/resnet_utils.pyhttps://github.com/tensorflow/models/blob/master/research/slim/nets/resnet_v1.py 1、TensorFlow中resnet_v1_50的用法 ...
基于VGG和Resnet50模型融合的图像分类项目源码解析
项目所使用的两个深度学习模型VGG和Resnet50均为卷积神经网络（CNN）架构，它们在图像识别和分类领域表现卓越，具有高度的影响力。 VGG模型是2014年牛津大学的研究者设计的一种深度卷积神经网络，以简洁的网络结构...
评论
添加红包

请填写红包祝福语或标题

个

红包个数最小为10个

元

红包金额最低5元

当前余额3.43元 前往充值 >
需支付：10.00元
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付元
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明：

1.余额是钱包充值的虚拟货币，按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载，可以购买VIP、付费专栏及课程。

余额充值