详解keras的卷积层conv2d model.summary()输出参数Param计算过程

详解keras的model.summary()输出参数Param计算过程 最难的是卷积层

1、代码产生conv_1层 他的 param 参数为:(通道数2*核宽2*核高2)+1)*卷积核数3=27

2、代码产生conv_2层 他的 param 参数为:(上层卷积核数3*核宽3*核高3)+1)*卷积核数24=672


    image = Input(shape=(5,5,通道数2),name="input_my")
    1、x = Conv2D(卷积核数3, kernel_size=(核宽2,核高2), strides=(1,1), activation='relu', name='conv_1')(image)
    2、x = Conv2D(24, kernel_size=(核宽3,核高3), strides=(1,1), activation='relu', name='conv_2')(x)
   


from keras import *
from keras.layers import Conv2D,Flatten,Dense
import numpy as np

def create_model():

#------------------------------------
    image = Input(shape=(5,5,2),name="input_my")
    x = Conv2D(3, kernel_size=(2,2), strides=(1,1), activation='relu', name='conv_1')(image)
    x = Conv2D(24, kernel_size=(3,3), strides=(1,1), activation='relu', name='conv_2')(x)
#-------------------------------------------

    output = Dense(1, activation='relu', name='output')(x)
    model = Model(inputs=image, outputs=output)
    return model

Layer (type)                 Output Shape              Param #   
=================================================================
input_my (InputLayer)        (None, 5, 5, 2)           0         
_________________________________________________________________
conv_1 (Conv2D)              (None, 4, 4, 3)           27        
_________________________________________________________________
conv_2 (Conv2D)              (None, 2, 2, 24)          672       
_________________________________________________________________
output (Dense)               (None, 2, 2, 1)           25        
=================================================================

你可能感兴趣的:(卷积,深度学习,神经网络)