File size: 4,799 Bytes
54c22e4 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 |
import torch
from torch import nn
import torch.nn.functional as F
from lib_v5 import spec_utils
class Conv2DBNActiv(nn.Module):
def __init__(self, nin, nout, ksize=3, stride=1, pad=1, dilation=1, activ=nn.ReLU):
super(Conv2DBNActiv, self).__init__()
self.conv = nn.Sequential(
nn.Conv2d(
nin, nout,
kernel_size=ksize,
stride=stride,
padding=pad,
dilation=dilation,
bias=False),
nn.BatchNorm2d(nout),
activ()
)
def __call__(self, x):
return self.conv(x)
class SeperableConv2DBNActiv(nn.Module):
def __init__(self, nin, nout, ksize=3, stride=1, pad=1, dilation=1, activ=nn.ReLU):
super(SeperableConv2DBNActiv, self).__init__()
self.conv = nn.Sequential(
nn.Conv2d(
nin, nin,
kernel_size=ksize,
stride=stride,
padding=pad,
dilation=dilation,
groups=nin,
bias=False),
nn.Conv2d(
nin, nout,
kernel_size=1,
bias=False),
nn.BatchNorm2d(nout),
activ()
)
def __call__(self, x):
return self.conv(x)
class Encoder(nn.Module):
def __init__(self, nin, nout, ksize=3, stride=1, pad=1, activ=nn.LeakyReLU):
super(Encoder, self).__init__()
self.conv1 = Conv2DBNActiv(nin, nout, ksize, 1, pad, activ=activ)
self.conv2 = Conv2DBNActiv(nout, nout, ksize, stride, pad, activ=activ)
def __call__(self, x):
skip = self.conv1(x)
h = self.conv2(skip)
return h, skip
class Decoder(nn.Module):
def __init__(self, nin, nout, ksize=3, stride=1, pad=1, activ=nn.ReLU, dropout=False):
super(Decoder, self).__init__()
self.conv = Conv2DBNActiv(nin, nout, ksize, 1, pad, activ=activ)
self.dropout = nn.Dropout2d(0.1) if dropout else None
def __call__(self, x, skip=None):
x = F.interpolate(x, scale_factor=2, mode='bilinear', align_corners=True)
if skip is not None:
skip = spec_utils.crop_center(skip, x)
x = torch.cat([x, skip], dim=1)
h = self.conv(x)
if self.dropout is not None:
h = self.dropout(h)
return h
class ASPPModule(nn.Module):
def __init__(self, nn_architecture, nin, nout, dilations=(4, 8, 16), activ=nn.ReLU):
super(ASPPModule, self).__init__()
self.conv1 = nn.Sequential(
nn.AdaptiveAvgPool2d((1, None)),
Conv2DBNActiv(nin, nin, 1, 1, 0, activ=activ)
)
self.nn_architecture = nn_architecture
self.six_layer = [129605]
self.seven_layer = [537238, 537227, 33966]
extra_conv = SeperableConv2DBNActiv(
nin, nin, 3, 1, dilations[2], dilations[2], activ=activ)
self.conv2 = Conv2DBNActiv(nin, nin, 1, 1, 0, activ=activ)
self.conv3 = SeperableConv2DBNActiv(
nin, nin, 3, 1, dilations[0], dilations[0], activ=activ)
self.conv4 = SeperableConv2DBNActiv(
nin, nin, 3, 1, dilations[1], dilations[1], activ=activ)
self.conv5 = SeperableConv2DBNActiv(
nin, nin, 3, 1, dilations[2], dilations[2], activ=activ)
if self.nn_architecture in self.six_layer:
self.conv6 = extra_conv
nin_x = 6
elif self.nn_architecture in self.seven_layer:
self.conv6 = extra_conv
self.conv7 = extra_conv
nin_x = 7
else:
nin_x = 5
self.bottleneck = nn.Sequential(
Conv2DBNActiv(nin * nin_x, nout, 1, 1, 0, activ=activ),
nn.Dropout2d(0.1)
)
def forward(self, x):
_, _, h, w = x.size()
feat1 = F.interpolate(self.conv1(x), size=(h, w), mode='bilinear', align_corners=True)
feat2 = self.conv2(x)
feat3 = self.conv3(x)
feat4 = self.conv4(x)
feat5 = self.conv5(x)
if self.nn_architecture in self.six_layer:
feat6 = self.conv6(x)
out = torch.cat((feat1, feat2, feat3, feat4, feat5, feat6), dim=1)
elif self.nn_architecture in self.seven_layer:
feat6 = self.conv6(x)
feat7 = self.conv7(x)
out = torch.cat((feat1, feat2, feat3, feat4, feat5, feat6, feat7), dim=1)
else:
out = torch.cat((feat1, feat2, feat3, feat4, feat5), dim=1)
bottle = self.bottleneck(out)
return bottle
|