# InceptionResNetv2 derived from:
# https://github.com/Cadene/pretrained-models.pytorch/blob/master/pretrainedmodels/models/inceptionresnetv2.py
import torch
import torch.nn
import thelper.nn
import thelper.nn.coordconv
[docs]class BasicConv2d(torch.nn.Module):
[docs] def __init__(self, in_planes, out_planes, kernel_size, stride, padding=0):
super().__init__()
self.conv = torch.nn.Conv2d(in_planes, out_planes,
kernel_size=kernel_size, stride=stride,
padding=padding, bias=False) # verify bias false
self.bn = torch.nn.BatchNorm2d(out_planes,
eps=0.001, # value found in tensorflow
momentum=0.1, # default pytorch value
affine=True)
self.relu = torch.nn.ReLU(inplace=False)
[docs] def forward(self, x):
x = self.conv(x)
x = self.bn(x)
x = self.relu(x)
return x
[docs]class Mixed_5b(torch.nn.Module):
[docs] def __init__(self):
super().__init__()
self.branch0 = BasicConv2d(192, 96, kernel_size=1, stride=1)
self.branch1 = torch.nn.Sequential(
BasicConv2d(192, 48, kernel_size=1, stride=1),
BasicConv2d(48, 64, kernel_size=5, stride=1, padding=2)
)
self.branch2 = torch.nn.Sequential(
BasicConv2d(192, 64, kernel_size=1, stride=1),
BasicConv2d(64, 96, kernel_size=3, stride=1, padding=1),
BasicConv2d(96, 96, kernel_size=3, stride=1, padding=1)
)
self.branch3 = torch.nn.Sequential(
torch.nn.AvgPool2d(3, stride=1, padding=1, count_include_pad=False),
BasicConv2d(192, 64, kernel_size=1, stride=1)
)
[docs] def forward(self, x):
x0 = self.branch0(x)
x1 = self.branch1(x)
x2 = self.branch2(x)
x3 = self.branch3(x)
out = torch.cat((x0, x1, x2, x3), 1)
return out
[docs]class Block35(torch.nn.Module):
[docs] def __init__(self, scale=1.0):
super().__init__()
self.scale = scale
self.branch0 = BasicConv2d(320, 32, kernel_size=1, stride=1)
self.branch1 = torch.nn.Sequential(
BasicConv2d(320, 32, kernel_size=1, stride=1),
BasicConv2d(32, 32, kernel_size=3, stride=1, padding=1)
)
self.branch2 = torch.nn.Sequential(
BasicConv2d(320, 32, kernel_size=1, stride=1),
BasicConv2d(32, 48, kernel_size=3, stride=1, padding=1),
BasicConv2d(48, 64, kernel_size=3, stride=1, padding=1)
)
self.conv2d = torch.nn.Conv2d(128, 320, kernel_size=1, stride=1)
self.relu = torch.nn.ReLU(inplace=False)
[docs] def forward(self, x):
x0 = self.branch0(x)
x1 = self.branch1(x)
x2 = self.branch2(x)
out = torch.cat((x0, x1, x2), 1)
out = self.conv2d(out)
out = out * self.scale + x
out = self.relu(out)
return out
[docs]class Mixed_6a(torch.nn.Module):
[docs] def __init__(self):
super().__init__()
self.branch0 = BasicConv2d(320, 384, kernel_size=3, stride=2)
self.branch1 = torch.nn.Sequential(
BasicConv2d(320, 256, kernel_size=1, stride=1),
BasicConv2d(256, 256, kernel_size=3, stride=1, padding=1),
BasicConv2d(256, 384, kernel_size=3, stride=2)
)
self.branch2 = torch.nn.MaxPool2d(3, stride=2)
[docs] def forward(self, x):
x0 = self.branch0(x)
x1 = self.branch1(x)
x2 = self.branch2(x)
out = torch.cat((x0, x1, x2), 1)
return out
[docs]class Block17(torch.nn.Module):
[docs] def __init__(self, scale=1.0):
super().__init__()
self.scale = scale
self.branch0 = BasicConv2d(1088, 192, kernel_size=1, stride=1)
self.branch1 = torch.nn.Sequential(
BasicConv2d(1088, 128, kernel_size=1, stride=1),
BasicConv2d(128, 160, kernel_size=(1, 7), stride=1, padding=(0, 3)),
BasicConv2d(160, 192, kernel_size=(7, 1), stride=1, padding=(3, 0))
)
self.conv2d = torch.nn.Conv2d(384, 1088, kernel_size=1, stride=1)
self.relu = torch.nn.ReLU(inplace=False)
[docs] def forward(self, x):
x0 = self.branch0(x)
x1 = self.branch1(x)
out = torch.cat((x0, x1), 1)
out = self.conv2d(out)
out = out * self.scale + x
out = self.relu(out)
return out
[docs]class Mixed_7a(torch.nn.Module):
[docs] def __init__(self):
super().__init__()
self.branch0 = torch.nn.Sequential(
BasicConv2d(1088, 256, kernel_size=1, stride=1),
BasicConv2d(256, 384, kernel_size=3, stride=2)
)
self.branch1 = torch.nn.Sequential(
BasicConv2d(1088, 256, kernel_size=1, stride=1),
BasicConv2d(256, 288, kernel_size=3, stride=2)
)
self.branch2 = torch.nn.Sequential(
BasicConv2d(1088, 256, kernel_size=1, stride=1),
BasicConv2d(256, 288, kernel_size=3, stride=1, padding=1),
BasicConv2d(288, 320, kernel_size=3, stride=2)
)
self.branch3 = torch.nn.MaxPool2d(3, stride=2)
[docs] def forward(self, x):
x0 = self.branch0(x)
x1 = self.branch1(x)
x2 = self.branch2(x)
x3 = self.branch3(x)
out = torch.cat((x0, x1, x2, x3), 1)
return out
[docs]class Block8(torch.nn.Module):
[docs] def __init__(self, scale=1.0, noReLU=False):
super(Block8, self).__init__()
self.scale = scale
self.noReLU = noReLU
self.branch0 = BasicConv2d(2080, 192, kernel_size=1, stride=1)
self.branch1 = torch.nn.Sequential(
BasicConv2d(2080, 192, kernel_size=1, stride=1),
BasicConv2d(192, 224, kernel_size=(1, 3), stride=1, padding=(0, 1)),
BasicConv2d(224, 256, kernel_size=(3, 1), stride=1, padding=(1, 0))
)
self.conv2d = torch.nn.Conv2d(448, 2080, kernel_size=1, stride=1)
if not self.noReLU:
self.relu = torch.nn.ReLU(inplace=False)
[docs] def forward(self, x):
x0 = self.branch0(x)
x1 = self.branch1(x)
out = torch.cat((x0, x1), 1)
out = self.conv2d(out)
out = out * self.scale + x
if not self.noReLU:
out = self.relu(out)
return out
[docs]class InceptionResNetV2(thelper.nn.Module):
[docs] def __init__(self, task, input_channels=3):
# note: must always forward args to base class to keep backup
super().__init__(task, input_channels=input_channels)
self.conv2d_1a = BasicConv2d(input_channels, 32, kernel_size=3, stride=2)
self.conv2d_2a = BasicConv2d(32, 32, kernel_size=3, stride=1)
self.conv2d_2b = BasicConv2d(32, 64, kernel_size=3, stride=1, padding=1)
self.maxpool_3a = torch.nn.MaxPool2d(3, stride=2)
self.conv2d_3b = BasicConv2d(64, 80, kernel_size=1, stride=1)
self.conv2d_4a = BasicConv2d(80, 192, kernel_size=3, stride=1)
self.maxpool_5a = torch.nn.MaxPool2d(3, stride=2)
self.mixed_5b = Mixed_5b()
self.repeat = torch.nn.Sequential(
Block35(scale=0.17),
Block35(scale=0.17),
Block35(scale=0.17),
Block35(scale=0.17),
Block35(scale=0.17),
Block35(scale=0.17),
Block35(scale=0.17),
Block35(scale=0.17),
Block35(scale=0.17),
Block35(scale=0.17)
)
self.mixed_6a = Mixed_6a()
self.repeat_1 = torch.nn.Sequential(
Block17(scale=0.10),
Block17(scale=0.10),
Block17(scale=0.10),
Block17(scale=0.10),
Block17(scale=0.10),
Block17(scale=0.10),
Block17(scale=0.10),
Block17(scale=0.10),
Block17(scale=0.10),
Block17(scale=0.10),
Block17(scale=0.10),
Block17(scale=0.10),
Block17(scale=0.10),
Block17(scale=0.10),
Block17(scale=0.10),
Block17(scale=0.10),
Block17(scale=0.10),
Block17(scale=0.10),
Block17(scale=0.10),
Block17(scale=0.10)
)
self.mixed_7a = Mixed_7a()
self.repeat_2 = torch.nn.Sequential(
Block8(scale=0.20),
Block8(scale=0.20),
Block8(scale=0.20),
Block8(scale=0.20),
Block8(scale=0.20),
Block8(scale=0.20),
Block8(scale=0.20),
Block8(scale=0.20),
Block8(scale=0.20)
)
self.block8 = Block8(noReLU=True)
self.conv2d_7b = BasicConv2d(2080, 1536, kernel_size=1, stride=1)
self.avgpool_1a = torch.nn.AvgPool2d(8, count_include_pad=False)
self.last_linear = torch.nn.Linear(1536, 1000)
self.set_task(task)
[docs] def features(self, input):
x = self.conv2d_1a(input)
x = self.conv2d_2a(x)
x = self.conv2d_2b(x)
x = self.maxpool_3a(x)
x = self.conv2d_3b(x)
x = self.conv2d_4a(x)
x = self.maxpool_5a(x)
x = self.mixed_5b(x)
x = self.repeat(x)
x = self.mixed_6a(x)
x = self.repeat_1(x)
x = self.mixed_7a(x)
x = self.repeat_2(x)
x = self.block8(x)
x = self.conv2d_7b(x)
return x
[docs] def logits(self, features):
x = self.avgpool_1a(features)
x = x.view(x.size(0), -1)
x = self.last_linear(x)
return x
[docs] def forward(self, input):
x = self.features(input)
x = self.logits(x)
return x
[docs] def set_task(self, task):
assert isinstance(task, thelper.tasks.Classification), "missing impl for non-classif task type"
num_classes = len(task.class_names)
if self.last_linear.out_features != num_classes:
self.last_linear = torch.nn.Linear(1536, num_classes)
self.task = task