目录
3、对于input_channel和output_channel的理解:
5、代码2:查看input_channel和output_channel
一、conv2d官方文档
1、文档解释:
torch.nn.Conv2d(in_channels, out_channels, kernel_size, stride=1, padding=0, dilation=1, groups=1, bias=True, padding_mode='zeros')
in_channels (int) – 输入图像的通道数,彩色图像三通道
out_channels (int) – 卷积后产生的输出通道数多少
padding_mode (string, optional) – 控制以什么样的方式填充
'zeros'
,'reflect'
,'replicate'
or'circular'
. Default:'zeros'
groups (int, optional) – Number of blocked connections from input channels to output channels. Default: 1
bias (bool, optional) – If
True
, adds a learnable bias to the output. Default:True
2、官方给的图片解释:
No padding是padding=0;
No strides是strides=1
3、对于input_channel和output_channel的理解:
如果In_channel为1,就会有两个卷积核,输出两个叠加的输出
4、代码1:查看output
import torch
import torchvision
from torch import nn
from torch.nn import Conv2d
from torch.utils.data import DataLoader
dataset = torchvision.datasets.CIFAR10("../data",train = False,transform=torchvision.transforms.ToTensor(),download=True)
dataloader = DataLoader(dataset,batch_size=64)
#搭建简单神经网络
class Tudui(nn.Module):
def __init__(self):
super(Tudui,self).__init__()
self.conv1 = Conv2d(3,6,3,stride=1,padding=0)
def forward(self,x):#前向计算得出结果
x=self.conv1(x)
return x
tudui = Tudui()
print(tudui)
结果:
5、代码2:查看input_channel和output_channel
import torch
import torchvision
from torch import nn
from torch.nn import Conv2d
from torch.utils.data import DataLoader
dataset = torchvision.datasets.CIFAR10("../data",train = False,transform=torchvision.transforms.ToTensor(),download=True)
dataloader = DataLoader(dataset,batch_size=64)
#搭建简单神经网络
class Tudui(nn.Module):
def __init__(self):
super(Tudui,self).__init__()
self.conv1 = Conv2d(3,6,3,stride=1,padding=0)
def forward(self,x):#前向计算得出结果
x=self.conv1(x)
return x
tudui = Tudui()
for data in dataloader:
imgs,targets = data
output = tudui(imgs)#把数据放到神经网络中去
print(imgs.shape)#没有卷积前的大小
print(output.shape)#卷积后
部分结果:
6、代码3在tensorboard中打开:
import torch
import torchvision
from torch import nn
from torch.nn import Conv2d
from torch.utils.data import DataLoader
from torch.utils.tensorboard import SummaryWriter
dataset = torchvision.datasets.CIFAR10("../data",train = False,transform=torchvision.transforms.ToTensor(),download=True)
dataloader = DataLoader(dataset,batch_size=64)
#搭建简单神经网络
class Tudui(nn.Module):
def __init__(self):
super(Tudui,self).__init__()
self.conv1 = Conv2d(3,6,3,stride=1,padding=0)
def forward(self,x):#前向计算得出结果
x=self.conv1(x)
return x
tudui = Tudui()
writer = SummaryWriter("../logs")
step = 0
for data in dataloader:
imgs,targets = data
output = tudui(imgs)#把数据放到神经网络中去
print(imgs.shape)#没有卷积前的大小
print(output.shape)#卷积后
#torch.Size([64, 3, 32, 32])
writer.add_images("input",imgs,step)
#torch.Size([64, 6, 30, 30])
output = torch.reshape(output,(-1,3,30,30))
writer.add_images("ouput",output,step)
step=step+1
结果:(输出是输入的2倍)
二、最大池化的使用
1、官网介绍:
①各类池化:
②最大池化详情文档:
torch.nn.MaxPool2d(kernel_size, stride=None, padding=0, dilation=1, return_indices=False, ceil_mode=False)
Parameters
kernel_size – 取最大值的窗口(int or tuple)
stride – 横向和纵向的路径大小(注意:默认值=kernel_size大小)
padding – 填充
dilation – 空洞卷积
return_indices – if
True
, will return the max indices along with the outputs. Useful for torch.nn.MaxUnpool2d lateceil_mode – True-ceil模式,否则:floor模式(eg:2.31,floor-2,ceiling-3)
2、池化操作讲解:
①先设置3*3的池化核
②覆盖,取最大值(最大池化操作)
③移动,假设stride=默认值=3(kernel size大小)
此时已经超出区域:
a.ceil_mode=True,保留
b.ceil_mode=False,不保留(默认情况)
2、代码操作参数:
①ceil_mode=True
import torch
from torch.nn import MaxPool2d
from torch import nn
input = torch.tensor([[1,2,0,3,1],
[0,1,2,3,1],
[1,2,1,0,0],
[5,2,3,1,1],
[2,1,0,1,1]],dtype = torch.float32)
input = torch.reshape(input,(-1,1,5,5))#-1是模糊计算
#print(input.shape)
class Tudui(nn.Module):
def __init__(self):
super(Tudui,self).__init__()
self.maxpool = MaxPool2d(kernel_size=3,ceil_mode=True)
def forward(self,input):
output = self.maxpool(input)
return output
tudui = Tudui()
output= tudui(input)
print(output)
②输出:
③seil_mode=False:
import torch
from torch.nn import MaxPool2d
from torch import nn
input = torch.tensor([[1,2,0,3,1],
[0,1,2,3,1],
[1,2,1,0,0],
[5,2,3,1,1],
[2,1,0,1,1]],dtype = torch.float32)
input = torch.reshape(input,(-1,1,5,5))#-1是模糊计算
#print(input.shape)
class Tudui(nn.Module):
def __init__(self):
super(Tudui,self).__init__()
self.maxpool = MaxPool2d(kernel_size=3,ceil_mode=False)
def forward(self,input):
output = self.maxpool(input)
return output
tudui = Tudui()
output= tudui(input)
print(output)
3、池化的作用
①主要作用:保持输入的特征,但同时把数据量减小
②实践代码:
import torch
import torchvision
from torch.nn import MaxPool2d
from torch import nn
from torch.utils.data import DataLoader
from torch.utils.tensorboard import SummaryWriter
dataset = torchvision.datasets.CIFAR10("../data",train = False,download=True,transform=torchvision.transforms.ToTensor())
dataloader = DataLoader(dataset,batch_size=64)
class Tudui(nn.Module):
def __init__(self):
super(Tudui,self).__init__()
self.maxpool = MaxPool2d(kernel_size=3,ceil_mode=False)
def forward(self,input):
output = self.maxpool(input)
return output
tudui = Tudui()
writer = SummaryWriter("../logs_maxpool")
step = 0
for data in dataloader:
imgs,targets = data
writer.add_images("input",imgs,step)
output = tudui(imgs)
writer.add_images("output",output,step)
step =step+1
writer.close()