填充与步幅

padding & stride

复制代码from mxnet import autograd,nd from mxnet import gluon,init from mxnet.gluon import nn,loss as gloss from mxnet.gluon import data as gdata def comp_conv2d(conv2d,X): conv2d.initialize() # (样本，通道，高，宽) X = X.reshape((1,1)+X.shape) #print(X.shape) Y = conv2d(X) return Y conv2d = nn.Conv2D(1,kernel_size=(3,3),padding=1) X = nd.random.uniform(shape=(8,8)) #print(X) #print(comp_conv2d(conv2d,X).shape) conv2d = nn.Conv2D(1,kernel_size=(5,3),padding=(2,1)) #print(comp_conv2d(conv2d,X).shape) conv2d = nn.Conv2D(1,kernel_size=3,padding=1,strides=2) print(comp_conv2d(conv2d,X).shape) conv2d = nn.Conv2D(1,kernel_size=(3,5),padding=(0,1),strides=(3,4)) print(comp_conv2d(conv2d,X).shape)

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
from mxnet import autograd,nd
from mxnet import gluon,init
from mxnet.gluon import nn,loss as gloss
from mxnet.gluon import data as gdata
def comp_conv2d(conv2d,X):
conv2d.initialize()
# (样本，通道，高，宽)
X = X.reshape((1,1)+X.shape)
#print(X.shape)
Y = conv2d(X)
return Y
conv2d = nn.Conv2D(1,kernel_size=(3,3),padding=1)
X = nd.random.uniform(shape=(8,8))
#print(X)
#print(comp_conv2d(conv2d,X).shape)

conv2d = nn.Conv2D(1,kernel_size=(5,3),padding=(2,1))
#print(comp_conv2d(conv2d,X).shape)

conv2d = nn.Conv2D(1,kernel_size=3,padding=1,strides=2)
print(comp_conv2d(conv2d,X).shape)
conv2d = nn.Conv2D(1,kernel_size=(3,5),padding=(0,1),strides=(3,4))
print(comp_conv2d(conv2d,X).shape)