File size: 13,039 Bytes
dae5c90 | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 | import torch
import torch.nn as nn
import torch.utils.model_zoo as model_zoo
from itertools import chain
import torch.utils.checkpoint as cp
from math import log2
from timm.models.registry import register_model
from ..utils.utils import _Upsample, SpatialPyramidPooling, SeparableConv2d
__all__ = ['ResNet', 'resnet18', 'resnet34', 'resnet50', 'resnet101', 'resnet152', 'BasicBlock']
model_urls = {
'resnet18': 'https://download.pytorch.org/models/resnet18-5c106cde.pth',
'resnet34': 'https://download.pytorch.org/models/resnet34-333f7ec4.pth',
'resnet50': 'https://download.pytorch.org/models/resnet50-19c8e357.pth',
'resnet101': 'https://download.pytorch.org/models/resnet101-5d3b4d8f.pth',
'resnet152': 'https://download.pytorch.org/models/resnet152-b121ed2d.pth',
}
def conv3x3(in_planes, out_planes, stride=1, separable=False):
"""3x3 convolution with padding"""
conv_class = SeparableConv2d if separable else nn.Conv2d
return conv_class(in_planes, out_planes, kernel_size=3, stride=stride, padding=1, bias=False)
def _bn_function_factory(conv, norm, relu=None):
def bn_function(x):
x = conv(x)
if norm is not None:
x = norm(x)
if relu is not None:
x = relu(x)
return x
return bn_function
def do_efficient_fwd(block, x, efficient):
if efficient and x.requires_grad:
return cp.checkpoint(block, x)
else:
return block(x)
class BasicBlock(nn.Module):
expansion = 1
def __init__(self, inplanes, planes, stride=1, downsample=None, efficient=True, use_bn=True, deleting=False,
separable=False):
super(BasicBlock, self).__init__()
self.use_bn = use_bn
self.conv1 = conv3x3(inplanes, planes, stride, separable=separable)
self.bn1 = nn.BatchNorm2d(planes) if self.use_bn else None
self.relu = nn.ReLU(inplace=True)
self.conv2 = conv3x3(planes, planes, separable=separable)
self.bn2 = nn.BatchNorm2d(planes) if self.use_bn else None
self.downsample = downsample
self.stride = stride
self.efficient = efficient
self.deleting = deleting
def forward(self, x):
residual = x
if self.downsample is not None:
residual = self.downsample(x)
if self.deleting is False:
bn_1 = _bn_function_factory(self.conv1, self.bn1, self.relu)
bn_2 = _bn_function_factory(self.conv2, self.bn2)
out = do_efficient_fwd(bn_1, x, self.efficient)
out = do_efficient_fwd(bn_2, out, self.efficient)
else:
out = torch.zeros_like(residual)
out = out + residual
relu = self.relu(out)
# print(f'Basic Block memory: {torch.cuda.memory_allocated() // 2**20}')
return relu, out
class Bottleneck(nn.Module):
expansion = 4
def __init__(self, inplanes, planes, stride=1, downsample=None, efficient=True, use_bn=True, separable=False):
super(Bottleneck, self).__init__()
self.use_bn = use_bn
self.conv1 = nn.Conv2d(inplanes, planes, kernel_size=1, bias=False)
self.bn1 = nn.BatchNorm2d(planes) if self.use_bn else None
conv_class = SeparableConv2d if separable else nn.Conv2d
self.conv2 = conv_class(planes, planes, kernel_size=3, stride=stride,
padding=1, bias=False)
self.bn2 = nn.BatchNorm2d(planes) if self.use_bn else None
self.conv3 = nn.Conv2d(planes, planes * self.expansion, kernel_size=1, bias=False)
self.bn3 = nn.BatchNorm2d(planes * self.expansion) if self.use_bn else None
self.relu = nn.ReLU(inplace=False)
self.downsample = downsample
self.stride = stride
self.efficient = efficient
def forward(self, x):
residual = x
bn_1 = _bn_function_factory(self.conv1, self.bn1, self.relu)
bn_2 = _bn_function_factory(self.conv2, self.bn2, self.relu)
bn_3 = _bn_function_factory(self.conv3, self.bn3, self.relu)
out = do_efficient_fwd(bn_1, x, self.efficient)
out = do_efficient_fwd(bn_2, out, self.efficient)
out = do_efficient_fwd(bn_3, out, self.efficient)
if self.downsample is not None:
residual = self.downsample(x)
out = out + residual
relu = self.relu(out)
return relu, out
class ResNet(nn.Module):
def __init__(self, block, layers, *, num_features=128, k_up=3, efficient=False, use_bn=True,
spp_grids=(8, 4, 2, 1), spp_square_grid=False, spp_drop_rate=0.0,
upsample_skip=True, upsample_only_skip=False,
detach_upsample_skips=(), detach_upsample_in=False,
target_size=None, output_stride=4, mean=(73.1584, 82.9090, 72.3924),
std=(44.9149, 46.1529, 45.3192), scale=1, separable=False,
upsample_separable=False, **kwargs):
super(ResNet, self).__init__()
self.inplanes = 64
self.efficient = efficient
self.use_bn = use_bn
self.separable = separable
self.register_buffer('img_mean', torch.tensor(mean).view(1, -1, 1, 1))
self.register_buffer('img_std', torch.tensor(std).view(1, -1, 1, 1))
if scale != 1:
self.register_buffer('img_scale', torch.tensor(scale).view(1, -1, 1, 1).float())
self.detach_upsample_in = detach_upsample_in
self.conv1 = nn.Conv2d(3, 64, kernel_size=7, stride=2, padding=3,
bias=False)
self.bn1 = nn.BatchNorm2d(64) if self.use_bn else lambda x: x
self.relu = nn.ReLU(inplace=True)
self.maxpool = nn.MaxPool2d(kernel_size=3, stride=2, padding=1)
self.target_size = target_size
if self.target_size is not None:
h, w = target_size
target_sizes = [(h // 2 ** i, w // 2 ** i) for i in range(2, 6)]
else:
target_sizes = [None] * 4
upsamples = []
self.layer1 = self._make_layer(block, 64, layers[0])
upsamples += [
_Upsample(num_features, self.inplanes, num_features, use_bn=self.use_bn, k=k_up, use_skip=upsample_skip,
only_skip=upsample_only_skip, detach_skip=2 in detach_upsample_skips, fixed_size=target_sizes[0],
separable=upsample_separable)]
self.layer2 = self._make_layer(block, 128, layers[1], stride=2)
upsamples += [
_Upsample(num_features, self.inplanes, num_features, use_bn=self.use_bn, k=k_up, use_skip=upsample_skip,
only_skip=upsample_only_skip, detach_skip=1 in detach_upsample_skips, fixed_size=target_sizes[1],
separable=upsample_separable)]
self.layer3 = self._make_layer(block, 256, layers[2], stride=2)
upsamples += [
_Upsample(num_features, self.inplanes, num_features, use_bn=self.use_bn, k=k_up, use_skip=upsample_skip,
only_skip=upsample_only_skip, detach_skip=0 in detach_upsample_skips, fixed_size=target_sizes[2],
separable=upsample_separable)]
self.layer4 = self._make_layer(block, 512, layers[3], stride=2)
self.fine_tune = [self.conv1, self.maxpool, self.layer1, self.layer2, self.layer3, self.layer4]
if self.use_bn:
self.fine_tune += [self.bn1]
num_levels = 3
self.spp_size = kwargs.get('spp_size', num_features)
bt_size = self.spp_size
level_size = self.spp_size // num_levels
self.spp = SpatialPyramidPooling(self.inplanes, num_levels, bt_size=bt_size, level_size=level_size,
out_size=num_features, grids=spp_grids, square_grid=spp_square_grid,
bn_momentum=0.01 / 2, use_bn=self.use_bn, drop_rate=spp_drop_rate
, fixed_size=target_sizes[3])
num_up_remove = max(0, int(log2(output_stride) - 2))
self.upsample = nn.ModuleList(list(reversed(upsamples[num_up_remove:])))
self.random_init = [self.spp, self.upsample]
self.num_features = num_features
for m in self.modules():
if isinstance(m, nn.Conv2d):
nn.init.kaiming_normal_(m.weight, mode='fan_out', nonlinearity='relu')
elif isinstance(m, nn.BatchNorm2d):
nn.init.constant_(m.weight, 1)
nn.init.constant_(m.bias, 0)
def _make_layer(self, block, planes, blocks, stride=1):
downsample = None
if stride != 1 or self.inplanes != planes * block.expansion:
layers = [nn.Conv2d(self.inplanes, planes * block.expansion, kernel_size=1, stride=stride, bias=False)]
if self.use_bn:
layers += [nn.BatchNorm2d(planes * block.expansion)]
downsample = nn.Sequential(*layers)
layers = [block(self.inplanes, planes, stride, downsample, efficient=self.efficient, use_bn=self.use_bn,
separable=self.separable)]
self.inplanes = planes * block.expansion
for i in range(1, blocks):
layers += [block(self.inplanes, planes, efficient=self.efficient, use_bn=self.use_bn,
separable=self.separable)]
return nn.Sequential(*layers)
def random_init_params(self):
return chain(*[f.parameters() for f in self.random_init])
def fine_tune_params(self):
return chain(*[f.parameters() for f in self.fine_tune])
def forward_resblock(self, x, layers):
skip = None
for l in layers:
x = l(x)
if isinstance(x, tuple):
x, skip = x
return x, skip
def forward_down(self, image):
if hasattr(self, 'img_scale'):
image /= self.img_scale
image -= self.img_mean
image /= self.img_std
x = self.conv1(image)
x = self.bn1(x)
x = self.relu(x)
x = self.maxpool(x)
features = []
x, skip = self.forward_resblock(x, self.layer1)
features += [skip]
x, skip = self.forward_resblock(x, self.layer2)
features += [skip]
x, skip = self.forward_resblock(x, self.layer3)
features += [skip]
x, skip = self.forward_resblock(x, self.layer4)
features += [self.spp.forward(skip)]
return features
def forward_up(self, features):
features = features[::-1]
x = features[0]
if self.detach_upsample_in:
x = x.detach()
upsamples = []
for skip, up in zip(features[1:], self.upsample):
x = up(x, skip)
upsamples += [x]
return x, {'features': features, 'upsamples': upsamples}
def forward(self, image):
return self.forward_up(self.forward_down(image))
@register_model
def resnet18(pretrained=True, **kwargs):
"""Constructs a ResNet-18 model.
Args:
pretrained (bool): If True, returns a model pre-trained on ImageNet
"""
model = ResNet(BasicBlock, [2, 2, 2, 2], **kwargs)
if pretrained:
model.load_state_dict(model_zoo.load_url(model_urls['resnet18']), strict=False)
return model
@register_model
def resnet34(pretrained=True, **kwargs):
"""Constructs a ResNet-34 model.
Args:
pretrained (bool): If True, returns a model pre-trained on ImageNet
"""
model = ResNet(BasicBlock, [3, 4, 6, 3], **kwargs)
if pretrained:
model.load_state_dict(model_zoo.load_url(model_urls['resnet34']), strict=False)
return model
@register_model
def resnet50(pretrained=True, **kwargs):
"""Constructs a ResNet-50 model.
Args:
pretrained (bool): If True, returns a model pre-trained on ImageNet
"""
model = ResNet(Bottleneck, [3, 4, 6, 3], **kwargs)
if pretrained:
model.load_state_dict(model_zoo.load_url(model_urls['resnet50']), strict=False)
return model
@register_model
def resnet101(pretrained=True, **kwargs):
"""Constructs a ResNet-101 model.
Args:
pretrained (bool): If True, returns a model pre-trained on ImageNet
"""
model = ResNet(Bottleneck, [3, 4, 23, 3], **kwargs)
if pretrained:
model.load_state_dict(model_zoo.load_url(model_urls['resnet101']), strict=False)
return model
@register_model
def resnet152(pretrained=True, **kwargs):
"""Constructs a ResNet-152 model.
Args:
pretrained (bool): If True, returns a model pre-trained on ImageNet
"""
model = ResNet(Bottleneck, [3, 8, 36, 3], **kwargs)
if pretrained:
model.load_state_dict(model_zoo.load_url(model_urls['resnet152']), strict=False)
return model |