|
import torch |
|
import torch.nn as nn |
|
import torchvision.models as models |
|
|
|
|
|
class UpsamplingAdd(nn.Module): |
|
def __init__(self, in_channels: int, out_channels: int, scale_factor: int = 2): |
|
super().__init__() |
|
self.upsample_layer = nn.Sequential( |
|
nn.Upsample( |
|
scale_factor=scale_factor, mode="bilinear", align_corners=False |
|
), |
|
nn.Conv2d(in_channels, out_channels, |
|
kernel_size=1, padding=0, bias=False), |
|
nn.InstanceNorm2d(out_channels), |
|
) |
|
|
|
def forward(self, x: torch.Tensor, x_skip: torch.Tensor): |
|
|
|
x = self.upsample_layer(x) |
|
|
|
if x.shape[-1] != x_skip.shape[-1] or x.shape[-2] != x_skip.shape[-2]: |
|
x = nn.functional.interpolate( |
|
x, size=(x_skip.shape[-2], x_skip.shape[-1]), mode="bilinear" |
|
) |
|
|
|
return x + x_skip |
|
|
|
|
|
class SegmentationHead(nn.Module): |
|
def __init__(self, in_channels: int, n_classes: int, dropout_rate: float = 0.0): |
|
super(SegmentationHead, self).__init__() |
|
|
|
backbone = models.resnet18(pretrained=False, zero_init_residual=True) |
|
|
|
self.first_conv = nn.Conv2d( |
|
in_channels, 64, kernel_size=7, stride=2, padding=3, bias=False |
|
) |
|
self.bn1 = backbone.bn1 |
|
self.relu = backbone.relu |
|
|
|
self.layer1 = backbone.layer1 |
|
self.layer2 = backbone.layer2 |
|
self.layer3 = backbone.layer3 |
|
|
|
|
|
self.up3_skip = UpsamplingAdd( |
|
in_channels=256, out_channels=128, scale_factor=2) |
|
self.up2_skip = UpsamplingAdd( |
|
in_channels=128, out_channels=64, scale_factor=2) |
|
self.up1_skip = UpsamplingAdd( |
|
in_channels=64, out_channels=in_channels, scale_factor=2) |
|
|
|
|
|
self.dropout = nn.Dropout( |
|
dropout_rate) if dropout_rate > 0 else nn.Identity() |
|
|
|
self.segmentation_head = nn.Sequential( |
|
nn.Conv2d(in_channels, in_channels, |
|
kernel_size=3, padding=1, bias=False), |
|
nn.InstanceNorm2d(in_channels), |
|
nn.ReLU(inplace=True), |
|
self.dropout, |
|
nn.Conv2d(in_channels, n_classes, kernel_size=1, padding=0), |
|
) |
|
|
|
def forward(self, x: torch.Tensor): |
|
|
|
skip_x = {"1": x} |
|
x = self.first_conv(x) |
|
x = self.bn1(x) |
|
x = self.relu(x) |
|
x = self.dropout(x) |
|
|
|
|
|
x = self.layer1(x) |
|
skip_x["2"] = x |
|
x = self.dropout(x) |
|
|
|
x = self.layer2(x) |
|
skip_x["3"] = x |
|
x = self.dropout(x) |
|
|
|
|
|
x = self.layer3(x) |
|
x = self.dropout(x) |
|
|
|
|
|
x = self.up3_skip(x, skip_x["3"]) |
|
x = self.dropout(x) |
|
|
|
|
|
x = self.up2_skip(x, skip_x["2"]) |
|
x = self.dropout(x) |
|
|
|
|
|
x = self.up1_skip(x, skip_x["1"]) |
|
x = self.dropout(x) |
|
|
|
segmentation_output = self.segmentation_head(x) |
|
|
|
return segmentation_output |
|
|