深度学习论文: Making Convolutional Networks Shift-Invariant Again及其PyTorch实现
Making Convolutional Networks Shift-Invariant Again
PDF: https://arxiv.org/pdf/1904.11486.pdf PyTorch代码: https://github.com/shanglianlm0525/CvPytorch PyTorch代码: https://github.com/shanglianlm0525/PyTorch-Networks
1 概述
CNN中的downsample过程由于不满足采样定理，所以没法确保平移不变性。信号处理里面解决这样的问题是利用增大采样频率或者用抗混叠方法，前者在图像处理里面设置stride 1就可实现，但stride 1已经是极限，本文着重于使用抗混叠使得CNN重新具有平移不变性。
2 Anti-aliasing
抗混叠通过抗混叠滤波器消除混叠，即先用低通滤波器处理，然后再去采样，这样可以消除高频信号造成的不满足采样定理的情况。
PyTorch代码：
import torch
import torch.nn as nn
import torch.nn.functional as F
import numpy as np
def get_pad_layer(pad_type):
if(pad_type in ['refl','reflect']):
PadLayer = nn.ReflectionPad2d
elif(pad_type in ['repl','replicate']):
PadLayer = nn.ReplicationPad2d
elif(pad_type=='zero'):
PadLayer = nn.ZeroPad2d
else:
print('Pad type [%s] not recognized'%pad_type)
return PadLayer
class BlurPool2D(nn.Module):
"""
Making Convolutional Networks Shift-Invariant Again
https://arxiv.org/abs/1904.11486
"""
def __init__(self, channels, pad_type='reflect', filt_size=4, stride=2, pad_off=0):
super(BlurPool2D, self).__init__()
self.filt_size = filt_size
self.pad_off = pad_off
self.pad_sizes = [int(1.*(filt_size-1)/2), int(np.ceil(1.*(filt_size-1)/2)), int(1.*(filt_size-1)/2), int(np.ceil(1.*(filt_size-1)/2))]
self.pad_sizes = [pad_size+pad_off for pad_size in self.pad_sizes]
self.stride = stride
self.off = int((self.stride-1)/2.)
self.channels = channels
if(self.filt_size==1):
a = np.array([1.,])
elif(self.filt_size==2):
a = np.array([1., 1.])
elif(self.filt_size==3):
a = np.array([1., 2., 1.])
elif(self.filt_size==4):
a = np.array([1., 3., 3., 1.])
elif(self.filt_size==5):
a = np.array([1., 4., 6., 4., 1.])
elif(self.filt_size==6):
a = np.array([1., 5., 10., 10., 5., 1.])
elif(self.filt_size==7):
a = np.array([1., 6., 15., 20., 15., 6., 1.])
filt = torch.Tensor(a[:,None]*a[None,:])
filt = filt/torch.sum(filt)
self.register_buffer('filt', filt[None,None,:,:].repeat((self.channels,1,1,1)))
self.pad = get_pad_layer(pad_type)(self.pad_sizes)
def forward(self, inp):
if(self.filt_size==1):
if(self.pad_off==0):
return inp[:,:,::self.stride,::self.stride]
else:
return self.pad(inp)[:,:,::self.stride,::self.stride]
else:
return F.conv2d(self.pad(inp), self.filt, stride=self.stride, groups=inp.shape[1])
