resnet50网络结构_基于 Tensorflon 实现 Resnet50 并基于 CIFAR-10 数据训练

最新推荐文章于 2023-01-11 20:58:47 发布
最新推荐文章于 2023-01-11 20:58:47 发布
阅读量212 收藏
点赞数
文章标签： resnet50网络结构

05247621b306faea0f2376146a2795ac.png

基于 Tensorflon 实现 Resnet50 并基于 CIFAR-10 数据训练

了解 Resnet50 网络结构

引入了残差结构，解决了训练时因网络过深而发生梯度消失的情况有着更好的特征提取能力

9631879b7f4f96683dde22c0074cf4f5.png

比如下

1448d6c8dd21babc7f405df7e0b9d6bb.png

面这

基于 Tensorflow 实现 Resnet50

参考网络

caffe可视化版本

模型构建

def identity_block(inputs, kernal_size, filters):
    filters1, filters2, filters3 = filters
    x = tf.keras.layers.Conv2D(filters1, (1, 1), padding='same')(inputs)
    x = tf.keras.layers.BatchNormalization()(x)
    x = tf.keras.layers.Activation('relu')(x)

    x = tf.keras.layers.Conv2D(filters2, (1, 1), padding='same')(x)
    x = tf.keras.layers.BatchNormalization()(x)
    x = tf.keras.layers.Activation('relu')(x)

    x = tf.keras.layers.Conv2D(filters3, kernal_size, padding='same')(x)
    x = tf.keras.layers.BatchNormalization()(x)
    x = tf.keras.layers.Activation('relu')(x)

    x = tf.keras.layers.add([x, inputs])
    x = tf.keras.layers.Activation('relu')(x)

    return x


def conv_block(inputs, kernal_size, filters, strides=(2, 2)):
    filters1, filters2, filters3 = filters
    x = tf.keras.layers.Conv2D(filters1, (1, 1), strides=strides, padding='same')(inputs)
    x = tf.keras.layers.BatchNormalization()(x)
    x = tf.keras.layers.Activation('relu')(x)

    x = tf.keras.layers.Conv2D(filters2, kernal_size, strides=strides, padding='same')(x)
    x = tf.keras.layers.BatchNormalization()(x)
    x = tf.keras.layers.Activation('relu')(x)

    x = tf.keras.layers.Conv2D(filters3, (1, 1), strides=strides, padding='same')(x)
    x = tf.keras.layers.BatchNormalization()(x)
    x = tf.keras.layers.Activation('relu')(x)

    shortcut = tf.keras.layers.Conv2D(filters3, (1, 1), strides=strides, padding='same')(x)
    shortcut = tf.keras.layers.BatchNormalization()(shortcut)

    x = tf.keras.layers.add([x, shortcut])
    tf.keras.layers.Activation('relu')(x)

    return x


def resnet50(inputs, classes):
    "开始有7*7的"
    x = tf.keras.layers.Conv2D(64, (7, 7), strides=(2, 2), padding='same')(inputs)
    x = tf.keras.layers.BatchNormalization()(x)
    x = tf.keras.layers.Activation('relu')(x)

    x = tf.keras.layers.MaxPool2D((3, 3), strides=(2, 2))(x)
    x = conv_block(x, 3, [64, 64, 256])
    x = identity_block(x, 3, [64, 64, 256])
    x = identity_block(x, 3, [64, 64, 256])

    x = conv_block(x, 3, [128, 128, 512])
    x = identity_block(x, 3, [128, 128, 512])
    x = identity_block(x, 3, [128, 128, 512])
    x = identity_block(x, 3, [128, 128, 512])

    x = conv_block(x, 3, [256, 256, 1024])
    x = identity_block(x, 3, [256, 256, 1024])
    x = identity_block(x, 3, [256, 256, 1024])
    x = identity_block(x, 3, [256, 256, 1024])
    x = identity_block(x, 3, [256, 256, 1024])
    x = identity_block(x, 3, [256, 256, 1024])

    x = conv_block(x, 3, [512, 512, 2048])
    x = identity_block(x, 3, [512, 512, 2048])
    x = identity_block(x, 3, [512, 512, 2048])

    x = tf.keras.layers.GlobalAveragePooling2D()(x)
    x = tf.keras.layers.Dense(classes, activation='softmax')(x)

    return x
  
inputs = tf.keras.Input(shape=[32, 32, 3])
model = tf.keras.Model(inputs=inputs, outputs=resnet50(inputs, classes=10))
model.summary()

参数信息

Model: "model"
__________________________________________________________________________________________________
Layer (type)                    Output Shape         Param #     Connected to                     
==================================================================================================
input_1 (InputLayer)            [(None, 32, 32, 3)]  0                                            
__________________________________________________________________________________________________
conv2d (Conv2D)                 (None, 16, 16, 64)   9472        input_1[0][0]                    
__________________________________________________________________________________________________
batch_normalization (BatchNorma (None, 16, 16, 64)   256         conv2d[0][0]                     
__________________________________________________________________________________________________
activation (Activation)         (None, 16, 16, 64)   0           batch_normalization[0][0]        
__________________________________________________________________________________________________
max_pooling2d (MaxPooling2D)    (None, 7, 7, 64)     0           activation[0][0]                 
__________________________________________________________________________________________________
conv2d_1 (Conv2D)               (None, 4, 4, 64)     4160        max_pooling2d[0][0]              
__________________________________________________________________________________________________
batch_normalization_1 (BatchNor (None, 4, 4, 64)     256         conv2d_1[0][0]                   
__________________________________________________________________________________________________
activation_1 (Activation)       (None, 4, 4, 64)     0           batch_normalization_1[0][0]      
__________________________________________________________________________________________________
conv2d_2 (Conv2D)               (None, 2, 2, 64)     36928       activation_1[0][0]               
__________________________________________________________________________________________________
batch_normalization_2 (BatchNor (None, 2, 2, 64)     256         conv2d_2[0][0]                   
__________________________________________________________________________________________________
activation_2 (Activation)       (None, 2, 2, 64)     0           batch_normalization_2[0][0]      
__________________________________________________________________________________________________
conv2d_3 (Conv2D)               (None, 1, 1, 256)    16640       activation_2[0][0]               
__________________________________________________________________________________________________
batch_normalization_3 (BatchNor (None, 1, 1, 256)    1024        conv2d_3[0][0]                   
__________________________________________________________________________________________________
activation_3 (Activation)       (None, 1, 1, 256)    0           batch_normalization_3[0][0]      
__________________________________________________________________________________________________
conv2d_4 (Conv2D)               (None, 1, 1, 256)    65792       activation_3[0][0]               
__________________________________________________________________________________________________
batch_normalization_4 (BatchNor (None, 1, 1, 256)    1024        conv2d_4[0][0]                   
__________________________________________________________________________________________________
add (Add)                       (None, 1, 1, 256)    0           activation_3[0][0]               
                                                                 batch_normalization_4[0][0]      
__________________________________________________________________________________________________
conv2d_5 (Conv2D)               (None, 1, 1, 64)     16448       add[0][0]                        
__________________________________________________________________________________________________
batch_normalization_5 (BatchNor (None, 1, 1, 64)     256         conv2d_5[0][0]                   
__________________________________________________________________________________________________
activation_5 (Activation)       (None, 1, 1, 64)     0           batch_normalization_5[0][0]      
__________________________________________________________________________________________________
conv2d_6 (Conv2D)               (None, 1, 1, 64)     4160        activation_5[0][0]               
__________________________________________________________________________________________________
batch_normalization_6 (BatchNor (None, 1, 1, 64)     256         conv2d_6[0][0]                   
__________________________________________________________________________________________________
activation_6 (Activation)       (None, 1, 1, 64)     0           batch_normalization_6[0][0]      
__________________________________________________________________________________________________
conv2d_7 (Conv2D)               (None, 1, 1, 256)    147712      activation_6[0][0]               
__________________________________________________________________________________________________
batch_normalization_7 (BatchNor (None, 1, 1, 256)    1024        conv2d_7[0][0]                   
__________________________________________________________________________________________________
activation_7 (Activation)       (None, 1, 1, 256)    0           batch_normalization_7[0][0]      
__________________________________________________________________________________________________
add_1 (Add)                     (None, 1, 1, 256)    0           activation_7[0][0]               
                                                                 add[0][0]                        
__________________________________________________________________________________________________
activation_8 (Activation)       (None, 1, 1, 256)    0           add_1[0][0]                      
__________________________________________________________________________________________________
conv2d_8 (Conv2D)               (None, 1, 1, 64)     16448       activation_8[0][0]               
__________________________________________________________________________________________________
batch_normalization_8 (BatchNor (None, 1, 1, 64)     256         conv2d_8[0][0]                   
__________________________________________________________________________________________________
activation_9 (Activation)       (None, 1, 1, 64)     0           batch_normalization_8[0][0]      
__________________________________________________________________________________________________
conv2d_9 (Conv2D)               (None, 1, 1, 64)     4160        activation_9[0][0]               
__________________________________________________________________________________________________
batch_normalization_9 (BatchNor (None, 1, 1, 64)     256         conv2d_9[0][0]                   
__________________________________________________________________________________________________
activation_10 (Activation)      (None, 1, 1, 64)     0           batch_normalization_9[0][0]      
__________________________________________________________________________________________________
conv2d_10 (Conv2D)              (None, 1, 1, 256)    147712      activation_10[0][0]              
__________________________________________________________________________________________________
batch_normalization_10 (BatchNo (None, 1, 1, 256)    1024        conv2d_10[0][0]                  
__________________________________________________________________________________________________
activation_11 (Activation)      (None, 1, 1, 256)    0           batch_normalization_10[0][0]     
__________________________________________________________________________________________________
add_2 (Add)                     (None, 1, 1, 256)    0           activation_11[0][0]              
                                                                 activation_8[0][0]               
__________________________________________________________________________________________________
activation_12 (Activation)      (None, 1, 1, 256)    0           add_2[0][0]                      
__________________________________________________________________________________________________
conv2d_11 (Conv2D)              (None, 1, 1, 128)    32896       activation_12[0][0]              
__________________________________________________________________________________________________
batch_normalization_11 (BatchNo (None, 1, 1, 128)    512         conv2d_11[0][0]                  
__________________________________________________________________________________________________
activation_13 (Activation)      (None, 1, 1, 128)    0           batch_normalization_11[0][0]     
__________________________________________________________________________________________________
conv2d_12 (Conv2D)              (None, 1, 1, 128)    147584      activation_13[0][0]              
__________________________________________________________________________________________________
batch_normalization_12 (BatchNo (None, 1, 1, 128)    512         conv2d_12[0][0]                  
__________________________________________________________________________________________________
activation_14 (Activation)      (None, 1, 1, 128)    0           batch_normalization_12[0][0]     
__________________________________________________________________________________________________
conv2d_13 (Conv2D)              (None, 1, 1, 512)    66048       activation_14[0][0]              
__________________________________________________________________________________________________
batch_normalization_13 (BatchNo (None, 1, 1, 512)    2048        conv2d_13[0][0]                  
__________________________________________________________________________________________________
activation_15 (Activation)      (None, 1, 1, 512)    0           batch_normalization_13[0][0]     
__________________________________________________________________________________________________
conv2d_14 (Conv2D)              (None, 1, 1, 512)    262656      activation_15[0][0]              
__________________________________________________________________________________________________
batch_normalization_14 (BatchNo (None, 1, 1, 512)    2048        conv2d_14[0][0]                  
__________________________________________________________________________________________________
add_3 (Add)                     (None, 1, 1, 512)    0           activation_15[0][0]              
                                                                 batch_normalization_14[0][0]     
__________________________________________________________________________________________________
conv2d_15 (Conv2D)              (None, 1, 1, 128)    65664       add_3[0][0]                      
__________________________________________________________________________________________________
batch_normalization_15 (BatchNo (None, 1, 1, 128)    512         conv2d_15[0][0]                  
__________________________________________________________________________________________________
activation_17 (Activation)      (None, 1, 1, 128)    0           batch_normalization_15[0][0]     
__________________________________________________________________________________________________
conv2d_16 (Conv2D)              (None, 1, 1, 128)    16512       activation_17[0][0]              
__________________________________________________________________________________________________
batch_normalization_16 (BatchNo (None, 1, 1, 128)    512         conv2d_16[0][0]                  
__________________________________________________________________________________________________
activation_18 (Activation)      (None, 1, 1, 128)    0           batch_normalization_16[0][0]     
__________________________________________________________________________________________________
conv2d_17 (Conv2D)              (None, 1, 1, 512)    590336      activation_18[0][0]              
__________________________________________________________________________________________________
batch_normalization_17 (BatchNo (None, 1, 1, 512)    2048        conv2d_17[0][0]                  
__________________________________________________________________________________________________
activation_19 (Activation)      (None, 1, 1, 512)    0           batch_normalization_17[0][0]     
__________________________________________________________________________________________________
add_4 (Add)                     (None, 1, 1, 512)    0           activation_19[0][0]              
                                                                 add_3[0][0]                      
__________________________________________________________________________________________________
activation_20 (Activation)      (None, 1, 1, 512)    0           add_4[0][0]                      
__________________________________________________________________________________________________
conv2d_18 (Conv2D)              (None, 1, 1, 128)    65664       activation_20[0][0]              
__________________________________________________________________________________________________
batch_normalization_18 (BatchNo (None, 1, 1, 128)    512         conv2d_18[0][0]                  
__________________________________________________________________________________________________
activation_21 (Activation)      (None, 1, 1, 128)    0           batch_normalization_18[0][0]     
__________________________________________________________________________________________________
conv2d_19 (Conv2D)              (None, 1, 1, 128)    16512       activation_21[0][0]              
__________________________________________________________________________________________________
batch_normalization_19 (BatchNo (None, 1, 1, 128)    512         conv2d_19[0][0]                  
__________________________________________________________________________________________________
activation_22 (Activation)      (None, 1, 1, 128)    0           batch_normalization_19[0][0]     
__________________________________________________________________________________________________
conv2d_20 (Conv2D)              (None, 1, 1, 512)    590336      activation_22[0][0]              
__________________________________________________________________________________________________
batch_normalization_20 (BatchNo (None, 1, 1, 512)    2048        conv2d_20[0][0]                  
__________________________________________________________________________________________________
activation_23 (Activation)      (None, 1, 1, 512)    0           batch_normalization_20[0][0]     
__________________________________________________________________________________________________
add_5 (Add)                     (None, 1, 1, 512)    0           activation_23[0][0]              
                                                                 activation_20[0][0]              
__________________________________________________________________________________________________
activation_24 (Activation)      (None, 1, 1, 512)    0           add_5[0][0]                      
__________________________________________________________________________________________________
conv2d_21 (Conv2D)              (None, 1, 1, 128)    65664       activation_24[0][0]              
__________________________________________________________________________________________________
batch_normalization_21 (BatchNo (None, 1, 1, 128)    512         conv2d_21[0][0]                  
__________________________________________________________________________________________________
activation_25 (Activation)      (None, 1, 1, 128)    0           batch_normalization_21[0][0]     
__________________________________________________________________________________________________
conv2d_22 (Conv2D)              (None, 1, 1, 128)    16512       activation_25[0][0]              
__________________________________________________________________________________________________
batch_normalization_22 (BatchNo (None, 1, 1, 128)    512         conv2d_22[0][0]                  
__________________________________________________________________________________________________
activation_26 (Activation)      (None, 1, 1, 128)    0           batch_normalization_22[0][0]     
__________________________________________________________________________________________________
conv2d_23 (Conv2D)              (None, 1, 1, 512)    590336      activation_26[0][0]              
__________________________________________________________________________________________________
batch_normalization_23 (BatchNo (None, 1, 1, 512)    2048        conv2d_23[0][0]                  
__________________________________________________________________________________________________
activation_27 (Activation)      (None, 1, 1, 512)    0           batch_normalization_23[0][0]     
__________________________________________________________________________________________________
add_6 (Add)                     (None, 1, 1, 512)    0           activation_27[0][0]              
                                                                 activation_24[0][0]              
__________________________________________________________________________________________________
activation_28 (Activation)      (None, 1, 1, 512)    0           add_6[0][0]                      
__________________________________________________________________________________________________
conv2d_24 (Conv2D)              (None, 1, 1, 256)    131328      activation_28[0][0]              
__________________________________________________________________________________________________
batch_normalization_24 (BatchNo (None, 1, 1, 256)    1024        conv2d_24[0][0]                  
__________________________________________________________________________________________________
activation_29 (Activation)      (None, 1, 1, 256)    0           batch_normalization_24[0][0]     
__________________________________________________________________________________________________
conv2d_25 (Conv2D)              (None, 1, 1, 256)    590080      activation_29[0][0]              
__________________________________________________________________________________________________
batch_normalization_25 (BatchNo (None, 1, 1, 256)    1024        conv2d_25[0][0]                  
__________________________________________________________________________________________________
activation_30 (Activation)      (None, 1, 1, 256)    0           batch_normalization_25[0][0]     
__________________________________________________________________________________________________
conv2d_26 (Conv2D)              (None, 1, 1, 1024)   263168      activation_30[0][0]              
__________________________________________________________________________________________________
batch_normalization_26 (BatchNo (None, 1, 1, 1024)   4096        conv2d_26[0][0]                  
__________________________________________________________________________________________________
activation_31 (Activation)      (None, 1, 1, 1024)   0           batch_normalization_26[0][0]     
__________________________________________________________________________________________________
conv2d_27 (Conv2D)              (None, 1, 1, 1024)   1049600     activation_31[0][0]              
__________________________________________________________________________________________________
batch_normalization_27 (BatchNo (None, 1, 1, 1024)   4096        conv2d_27[0][0]                  
__________________________________________________________________________________________________
add_7 (Add)                     (None, 1, 1, 1024)   0           activation_31[0][0]              
                                                                 batch_normalization_27[0][0]     
__________________________________________________________________________________________________
conv2d_28 (Conv2D)              (None, 1, 1, 256)    262400      add_7[0][0]                      
__________________________________________________________________________________________________
batch_normalization_28 (BatchNo (None, 1, 1, 256)    1024        conv2d_28[0][0]                  
__________________________________________________________________________________________________
activation_33 (Activation)      (None, 1, 1, 256)    0           batch_normalization_28[0][0]     
__________________________________________________________________________________________________
conv2d_29 (Conv2D)              (None, 1, 1, 256)    65792       activation_33[0][0]              
__________________________________________________________________________________________________
batch_normalization_29 (BatchNo (None, 1, 1, 256)    1024        conv2d_29[0][0]                  
__________________________________________________________________________________________________
activation_34 (Activation)      (None, 1, 1, 256)    0           batch_normalization_29[0][0]     
__________________________________________________________________________________________________
conv2d_30 (Conv2D)              (None, 1, 1, 1024)   2360320     activation_34[0][0]              
__________________________________________________________________________________________________
batch_normalization_30 (BatchNo (None, 1, 1, 1024)   4096        conv2d_30[0][0]                  
__________________________________________________________________________________________________
activation_35 (Activation)      (None, 1, 1, 1024)   0           batch_normalization_30[0][0]     
__________________________________________________________________________________________________
add_8 (Add)                     (None, 1, 1, 1024)   0           activation_35[0][0]              
                                                                 add_7[0][0]                      
__________________________________________________________________________________________________
activation_36 (Activation)      (None, 1, 1, 1024)   0           add_8[0][0]                      
__________________________________________________________________________________________________
conv2d_31 (Conv2D)              (None, 1, 1, 256)    262400      activation_36[0][0]              
__________________________________________________________________________________________________
batch_normalization_31 (BatchNo (None, 1, 1, 256)    1024        conv2d_31[0][0]                  
__________________________________________________________________________________________________
activation_37 (Activation)      (None, 1, 1, 256)    0           batch_normalization_31[0][0]     
__________________________________________________________________________________________________
conv2d_32 (Conv2D)              (None, 1, 1, 256)    65792       activation_37[0][0]              
__________________________________________________________________________________________________
batch_normalization_32 (BatchNo (None, 1, 1, 256)    1024        conv2d_32[0][0]                  
__________________________________________________________________________________________________
activation_38 (Activation)      (None, 1, 1, 256)    0           batch_normalization_32[0][0]     
__________________________________________________________________________________________________
conv2d_33 (Conv2D)              (None, 1, 1, 1024)   2360320     activation_38[0][0]              
__________________________________________________________________________________________________
batch_normalization_33 (BatchNo (None, 1, 1, 1024)   4096        conv2d_33[0][0]                  
__________________________________________________________________________________________________
activation_39 (Activation)      (None, 1, 1, 1024)   0           batch_normalization_33[0][0]     
__________________________________________________________________________________________________
add_9 (Add)                     (None, 1, 1, 1024)   0           activation_39[0][0]              
                                                                 activation_36[0][0]              
__________________________________________________________________________________________________
activation_40 (Activation)      (None, 1, 1, 1024)   0           add_9[0][0]                      
__________________________________________________________________________________________________
conv2d_34 (Conv2D)              (None, 1, 1, 256)    262400      activation_40[0][0]              
__________________________________________________________________________________________________
batch_normalization_34 (BatchNo (None, 1, 1, 256)    1024        conv2d_34[0][0]                  
__________________________________________________________________________________________________
activation_41 (Activation)      (None, 1, 1, 256)    0           batch_normalization_34[0][0]     
__________________________________________________________________________________________________
conv2d_35 (Conv2D)              (None, 1, 1, 256)    65792       activation_41[0][0]              
__________________________________________________________________________________________________
batch_normalization_35 (BatchNo (None, 1, 1, 256)    1024        conv2d_35[0][0]                  
__________________________________________________________________________________________________
activation_42 (Activation)      (None, 1, 1, 256)    0           batch_normalization_35[0][0]     
__________________________________________________________________________________________________
conv2d_36 (Conv2D)              (None, 1, 1, 1024)   2360320     activation_42[0][0]              
__________________________________________________________________________________________________
batch_normalization_36 (BatchNo (None, 1, 1, 1024)   4096        conv2d_36[0][0]                  
__________________________________________________________________________________________________
activation_43 (Activation)      (None, 1, 1, 1024)   0           batch_normalization_36[0][0]     
__________________________________________________________________________________________________
add_10 (Add)                    (None, 1, 1, 1024)   0           activation_43[0][0]              
                                                                 activation_40[0][0]              
__________________________________________________________________________________________________
activation_44 (Activation)      (None, 1, 1, 1024)   0           add_10[0][0]                     
__________________________________________________________________________________________________
conv2d_37 (Conv2D)              (None, 1, 1, 256)    262400      activation_44[0][0]              
__________________________________________________________________________________________________
batch_normalization_37 (BatchNo (None, 1, 1, 256)    1024        conv2d_37[0][0]                  
__________________________________________________________________________________________________
activation_45 (Activation)      (None, 1, 1, 256)    0           batch_normalization_37[0][0]     
__________________________________________________________________________________________________
conv2d_38 (Conv2D)              (None, 1, 1, 256)    65792       activation_45[0][0]              
__________________________________________________________________________________________________
batch_normalization_38 (BatchNo (None, 1, 1, 256)    1024        conv2d_38[0][0]                  
__________________________________________________________________________________________________
activation_46 (Activation)      (None, 1, 1, 256)    0           batch_normalization_38[0][0]     
__________________________________________________________________________________________________
conv2d_39 (Conv2D)              (None, 1, 1, 1024)   2360320     activation_46[0][0]              
__________________________________________________________________________________________________
batch_normalization_39 (BatchNo (None, 1, 1, 1024)   4096        conv2d_39[0][0]                  
__________________________________________________________________________________________________
activation_47 (Activation)      (None, 1, 1, 1024)   0           batch_normalization_39[0][0]     
__________________________________________________________________________________________________
add_11 (Add)                    (None, 1, 1, 1024)   0           activation_47[0][0]              
                                                                 activation_44[0][0]              
__________________________________________________________________________________________________
activation_48 (Activation)      (None, 1, 1, 1024)   0           add_11[0][0]                     
__________________________________________________________________________________________________
conv2d_40 (Conv2D)              (None, 1, 1, 256)    262400      activation_48[0][0]              
__________________________________________________________________________________________________
batch_normalization_40 (BatchNo (None, 1, 1, 256)    1024        conv2d_40[0][0]                  
__________________________________________________________________________________________________
activation_49 (Activation)      (None, 1, 1, 256)    0           batch_normalization_40[0][0]     
__________________________________________________________________________________________________
conv2d_41 (Conv2D)              (None, 1, 1, 256)    65792       activation_49[0][0]              
__________________________________________________________________________________________________
batch_normalization_41 (BatchNo (None, 1, 1, 256)    1024        conv2d_41[0][0]                  
__________________________________________________________________________________________________
activation_50 (Activation)      (None, 1, 1, 256)    0           batch_normalization_41[0][0]     
__________________________________________________________________________________________________
conv2d_42 (Conv2D)              (None, 1, 1, 1024)   2360320     activation_50[0][0]              
__________________________________________________________________________________________________
batch_normalization_42 (BatchNo (None, 1, 1, 1024)   4096        conv2d_42[0][0]                  
__________________________________________________________________________________________________
activation_51 (Activation)      (None, 1, 1, 1024)   0           batch_normalization_42[0][0]     
__________________________________________________________________________________________________
add_12 (Add)                    (None, 1, 1, 1024)   0           activation_51[0][0]              
                                                                 activation_48[0][0]              
__________________________________________________________________________________________________
activation_52 (Activation)      (None, 1, 1, 1024)   0           add_12[0][0]                     
__________________________________________________________________________________________________
conv2d_43 (Conv2D)              (None, 1, 1, 512)    524800      activation_52[0][0]              
__________________________________________________________________________________________________
batch_normalization_43 (BatchNo (None, 1, 1, 512)    2048        conv2d_43[0][0]                  
__________________________________________________________________________________________________
activation_53 (Activation)      (None, 1, 1, 512)    0           batch_normalization_43[0][0]     
__________________________________________________________________________________________________
conv2d_44 (Conv2D)              (None, 1, 1, 512)    2359808     activation_53[0][0]              
__________________________________________________________________________________________________
batch_normalization_44 (BatchNo (None, 1, 1, 512)    2048        conv2d_44[0][0]                  
__________________________________________________________________________________________________
activation_54 (Activation)      (None, 1, 1, 512)    0           batch_normalization_44[0][0]     
__________________________________________________________________________________________________
conv2d_45 (Conv2D)              (None, 1, 1, 2048)   1050624     activation_54[0][0]              
__________________________________________________________________________________________________
batch_normalization_45 (BatchNo (None, 1, 1, 2048)   8192        conv2d_45[0][0]                  
__________________________________________________________________________________________________
activation_55 (Activation)      (None, 1, 1, 2048)   0           batch_normalization_45[0][0]     
__________________________________________________________________________________________________
conv2d_46 (Conv2D)              (None, 1, 1, 2048)   4196352     activation_55[0][0]              
__________________________________________________________________________________________________
batch_normalization_46 (BatchNo (None, 1, 1, 2048)   8192        conv2d_46[0][0]                  
__________________________________________________________________________________________________
add_13 (Add)                    (None, 1, 1, 2048)   0           activation_55[0][0]              
                                                                 batch_normalization_46[0][0]     
__________________________________________________________________________________________________
conv2d_47 (Conv2D)              (None, 1, 1, 512)    1049088     add_13[0][0]                     
__________________________________________________________________________________________________
batch_normalization_47 (BatchNo (None, 1, 1, 512)    2048        conv2d_47[0][0]                  
__________________________________________________________________________________________________
activation_57 (Activation)      (None, 1, 1, 512)    0           batch_normalization_47[0][0]     
__________________________________________________________________________________________________
conv2d_48 (Conv2D)              (None, 1, 1, 512)    262656      activation_57[0][0]              
__________________________________________________________________________________________________
batch_normalization_48 (BatchNo (None, 1, 1, 512)    2048        conv2d_48[0][0]                  
__________________________________________________________________________________________________
activation_58 (Activation)      (None, 1, 1, 512)    0           batch_normalization_48[0][0]     
__________________________________________________________________________________________________
conv2d_49 (Conv2D)              (None, 1, 1, 2048)   9439232     activation_58[0][0]              
__________________________________________________________________________________________________
batch_normalization_49 (BatchNo (None, 1, 1, 2048)   8192        conv2d_49[0][0]                  
__________________________________________________________________________________________________
activation_59 (Activation)      (None, 1, 1, 2048)   0           batch_normalization_49[0][0]     
__________________________________________________________________________________________________
add_14 (Add)                    (None, 1, 1, 2048)   0           activation_59[0][0]              
                                                                 add_13[0][0]                     
__________________________________________________________________________________________________
activation_60 (Activation)      (None, 1, 1, 2048)   0           add_14[0][0]                     
__________________________________________________________________________________________________
conv2d_50 (Conv2D)              (None, 1, 1, 512)    1049088     activation_60[0][0]              
__________________________________________________________________________________________________
batch_normalization_50 (BatchNo (None, 1, 1, 512)    2048        conv2d_50[0][0]                  
__________________________________________________________________________________________________
activation_61 (Activation)      (None, 1, 1, 512)    0           batch_normalization_50[0][0]     
__________________________________________________________________________________________________
conv2d_51 (Conv2D)              (None, 1, 1, 512)    262656      activation_61[0][0]              
__________________________________________________________________________________________________
batch_normalization_51 (BatchNo (None, 1, 1, 512)    2048        conv2d_51[0][0]                  
__________________________________________________________________________________________________
activation_62 (Activation)      (None, 1, 1, 512)    0           batch_normalization_51[0][0]     
__________________________________________________________________________________________________
conv2d_52 (Conv2D)              (None, 1, 1, 2048)   9439232     activation_62[0][0]              
__________________________________________________________________________________________________
batch_normalization_52 (BatchNo (None, 1, 1, 2048)   8192        conv2d_52[0][0]                  
__________________________________________________________________________________________________
activation_63 (Activation)      (None, 1, 1, 2048)   0           batch_normalization_52[0][0]     
__________________________________________________________________________________________________
add_15 (Add)                    (None, 1, 1, 2048)   0           activation_63[0][0]              
                                                                 activation_60[0][0]              
__________________________________________________________________________________________________
activation_64 (Activation)      (None, 1, 1, 2048)   0           add_15[0][0]                     
__________________________________________________________________________________________________
global_average_pooling2d (Globa (None, 2048)         0           activation_64[0][0]              
__________________________________________________________________________________________________
dense (Dense)                   (None, 10)           20490       global_average_pooling2d[0][0]   
==================================================================================================
Total params: 48,233,354
Trainable params: 48,180,234
Non-trainable params: 53,120
__________________________________________________________________________________________________

Process finished with exit code 0

数据准备

(x_train, y_train), (x_test, y_test) = tf.keras.datasets.cifar10.load_data()
x_train = x_train / 255.0
y_train = tf.keras.utils.to_categorical(y_train, 10)

模型训练

model.compile(loss='categorical_crossentropy', optimizer='adam', metrics=['categorical_accuracy', 'Recall', 'AUC'])
model.save('restnet50_cifar10')

模型应用

img = cv2.imread('catpng', 1)
img = np.expand_dims(img, 0)
model = tf.keras.modelsload_model('restnet50_cifar10.h5')
pred = model.predict(img)
print(pred)
weixin_39703468
关注
  • 0
    点赞
  • 踩
  • 0
    收藏
    觉得还不错? 一键收藏
  • 0
    评论
基于Keras+Resnet_v1对cifar-10进行图像分类
05-17
总的来说，这个项目展示了如何利用Keras和ResNet_v1来处理图像分类问题，并通过数据增强和适当的模型结构设计，提高了在CIFAR-10数据集上的分类精度。这种方法对于其他小型图像识别任务也有借鉴意义。通过深入理解...
resnet50_coco_best_v2.0.1.h5
10-22
在实际应用中，`resnet50_coco_best_v2.0.1.h5` 可能被用作基础模型，在其上添加额外的层来适应特定的任务，比如添加一个区域提议网络（Region Proposal Network, RPN）来生成候选框，再配合分类和回归层实现目标...
ResNet50 使用cifar-10数据集训练，出现了训练精度高，验证精度仅有75%的问题。还请各位同学不吝赐教！
rcl1245804161的博客
08-07 2772
ResNet50 使用cifar-10数据集训练，出现了训练精度高，验证精度仅有75%的问题。还请各位同学不吝赐教！ 由于是第一次在电脑上跑实验，代码是从TensorFlow2.0卷积神经网络实战上的代码，经过自己调试之后程序可以正常执行，但是结果不是很好，还请各位道友不吝赐教！ 下面抛出代码： 获取数据集的代码： import pickle import numpy as np import os from PIL import Image import tensorflow as tf def i
tensorflow 在cifar10上训练resnet50
What's past is prologue
10-28 5933
训练环境：windows10+python3.6.5+cuda-v9.0+tensorflow_gpu-1.10.0 目录   1、CIFAR-10介绍 2、resnet50网络介绍 3、使用TensorFlow Slim微调模型训练 3.1、下载TF-slim源码 3.2、下载cifar10数据并转换格式为trconf 3.3、下载resnet50模型 3.4、训练resnet...
实践 | 基于ResNet-50实现CIFAR10数据集分类
jiangti_ng的博客
05-24 3088
dog测成airplane，厉害了
使用ResNet50实现CIFAR10数据集的训练
走出那座山
01-11 2145
本文将实现使用Resnet模型进行数据集进行模型的训练，并且将训练的模型效果进行有用信息的记录和将在训练过程当中训练效果的ACC最优的模型进行保存。
pytorch利用resnet50实现cifar10准确率到95%以上
热门推荐
scarecrow_sun的博客
08-14 1万+
目录前言代码 前言 因为课程需要，老师要求使用resnet18或者resnet50将cifar10训练到精度到达95%，试过了网上其他很多方法，发现精度最高的是在预处理化的时候，图片resize到32,32，并且padding=4进行填充，并且优化器用SGD，准确率最高，但是准确率最高也是到88.8%。后来和上课的同学讨论，发现先将图片resize到(224,224)再进行翻转等操作可以稳定实现超过95%的准确率。 代码 废话不多说，直接上代码 注：因为CPU训练速度太慢了，尤其是将图片拉到224,224这
利用ResNet-50训练cifar-10数据集，精度达96%。训练记录（基于Pytorch）。
weixin_45264593的博客
08-03 2237
提示：文章写完后，目录可以自动生成，如何生成可参考右边的帮助文档 利用ResNet-50训练cifar-10数据集，精度达96%。训练记录（基于Pytorch）。前言一、这里我就简单概括一下。（我也想写具体点，但是没做详细的记录，cao!）1.数据增强2.模型结构二、代码1.数据处理2.训练3.配置文件（Configuration.ini）4.模型总结 前言 我在做基于卷积神经网路的注意力机制，这两天在搞自己的算法并与相关模型进行融合，首先就是ResNet-50模型。但在此之前，需要将基础模型训练好。于
卷积神经网络ResNet50训练CIFAR-100图像分类Pytorch实现
07-03
# 卷积神经网络ResNet50训练CIFAR-100图像分类Pytorch实现 1. 使用pytorch调用CIFAR-100数据集，首次训练自动下载； 2. 包含训练代码，调用resnet50模型进行训练，使用交叉熵损失和SGD优化器； 3. 包含训练了50 ...
resnet50_1000_imagenet_classifier.dnn.bz2
09-19
1000级别分类网络模型，采用50层的网络层，对大部分物体能检测出来，resnet50_1000_imagenet_classifier.dnn.bz2
tensorflow resnet源码 cifar10测试实战
09-28
ResNet源码，然后基于resnet实现在cifar10下的实战测试，注释齐全，特别详细，都能看懂。
利用pytorch实现对CIFAR-10数据集的分类
09-18
今天小编就为大家分享一篇利用pytorch实现对CIFAR-10数据集的分类，具有很好的参考价值，希望对大家有所帮助。一起跟随小编过来看看吧
基于ResNet50的CIFAR10分类
m0_51683462的博客
08-04 3942
基于ResNer50的CIFAR10分类，最优 acc 达到了91.14%
resnet50训练cifar10，请各位高手指正
masbbx123的博客
05-15 8820
使用resnet50从头训练cifar10，最终结果只有84%左右，貌似和 论文差很多，请各位高手指正。 首先加入cifar10的数据结构代码： import cifar10,cifar10_input import tensorflow as tf import numpy as np import time #max_steps = 100000 max_steps = 100 da...
PyTorch实现ResNet50,训练数据：CIFAR-10
weixin_36979214的博客
09-30 5378
ResNet网络结构 基本块 ResNet50代码实现 resnet50.py import torch import torch.nn as nn from torch.nn import functional as F class ResNet50BasicBlock(nn.Module): def __init__(self, in_channel, outs, kernerl_size, stride, padding): super(ResNet50BasicBl
[ 图像分类 ] 经典网络模型实例—— CIFAR10 + ResNet50 详解与复现
A_John 的博客
05-13 2900
[ 图像分类 ] 经典网络模型实例—— CIFAR10 + ResNet50 详解与复现 CIFAR10 + ResNet50 图像分类实例； 准备工作； 库文件准备； 命令行接口； 加载数据集； 加载模型； 定义损失函数与优化器； 打印模型； 模型训练与测试； 模型训练； 模型测试； 训练过程数据分析； 保存模型； 保存训练参数； 绘制 Loss-Acc 曲线； 主函数； CIFAR10 + ResNet50 完整代码；...
Pytorch实现Resnet训练CIFAR10数据集（完整代码，可进一步优化）
weixin_41823298的博客
09-21 1万+
Pytorch实现Resnet训练CIFAR10数据集 首先之前有写Pytorch的入门教程博客如果没有安装pytorch具体可转链接 废话不多说，直接上代码 这个代码使用CUDA 训练，如果不想使用GPU，可以将device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')注释掉，并且把所有后面用到的将tensor转到GPU的代码一起.to(device)删掉 import torch import torch.nn as nn
使用ResNet50， 对CIFAR-10进行分类
qq_42642142的博客
09-09 3156
使用ResNet50神经网络框架， 对CIFAR-10进行分类 训练集 50000， 测试集 10000 图像大小 32*32 彩色 10个分类： ariplane, automobile, bird, cat, deer, dog, frog, horse, ship, truck import torch import torch.nn as nn import torch.optim as optim import torchvision import torchvision.datasets a
用fastai ResNet50训练CIFAR10，85%准确度
weixin_30387799的博客
05-25 527
版权声明：本文为博主原创文章，欢迎转载，并请注明出处。联系方式：460356155@qq.com Fastai是在pytorch上封装的深度学习框架，效果出众，以下是训练CIFAR10的过程。 导入库 from fastai import * from fastai.vision import * from fastai.callbacks import CSVLog...
搭建resnet50网络结构训练cifar10
最新发布
10-23
如果要使用ResNet50网络结构训练CIFAR-10数据集，可以按照以下步骤进行： 1. 使用PyTorch调用CIFAR-10数据集，首次训练自动下载； 2. 搭建ResNet50网络结构，可以参考引用中的代码，将网络结构中的输出层改为10，以...
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