Spaces:
Running
Running
File size: 6,963 Bytes
c9baa67 |
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 |
"""
This code is adapted from: https://github.com/wielandbrendel/bag-of-local-features-models
"""
import torch.nn as nn
import math
import torch
from collections import OrderedDict
from torch.utils import model_zoo
from .normalizer import Normalizer
import os
dir_path = os.path.dirname(os.path.realpath(__file__))
__all__ = ['bagnet9', 'bagnet17', 'bagnet33']
model_urls = {
'bagnet9': 'https://bitbucket.org/wielandbrendel/bag-of-feature-pretrained-models/raw/249e8fa82c0913623a807d9d35eeab9da7dcc2a8/bagnet8-34f4ccd2.pth.tar',
'bagnet17': 'https://bitbucket.org/wielandbrendel/bag-of-feature-pretrained-models/raw/249e8fa82c0913623a807d9d35eeab9da7dcc2a8/bagnet16-105524de.pth.tar',
'bagnet33': 'https://bitbucket.org/wielandbrendel/bag-of-feature-pretrained-models/raw/249e8fa82c0913623a807d9d35eeab9da7dcc2a8/bagnet32-2ddd53ed.pth.tar',
}
class Bottleneck(nn.Module):
expansion = 4
def __init__(self, inplanes, planes, stride=1, downsample=None, kernel_size=1):
super(Bottleneck, self).__init__()
# print('Creating bottleneck with kernel size {} and stride {} with padding {}'.format(kernel_size, stride, (kernel_size - 1) // 2))
self.conv1 = nn.Conv2d(inplanes, planes, kernel_size=1, bias=False)
self.bn1 = nn.BatchNorm2d(planes)
self.conv2 = nn.Conv2d(planes, planes, kernel_size=kernel_size, stride=stride,
padding=0, bias=False) # changed padding from (kernel_size - 1) // 2
self.bn2 = nn.BatchNorm2d(planes)
self.conv3 = nn.Conv2d(planes, planes * 4, kernel_size=1, bias=False)
self.bn3 = nn.BatchNorm2d(planes * 4)
self.relu = nn.ReLU(inplace=True)
self.downsample = downsample
self.stride = stride
def forward(self, x, **kwargs):
residual = x
out = self.conv1(x)
out = self.bn1(out)
out = self.relu(out)
out = self.conv2(out)
out = self.bn2(out)
out = self.relu(out)
out = self.conv3(out)
out = self.bn3(out)
if self.downsample is not None:
residual = self.downsample(x)
if residual.size(-1) != out.size(-1):
diff = residual.size(-1) - out.size(-1)
residual = residual[:,:,:-diff,:-diff]
out += residual
out = self.relu(out)
return out
class BagNet(nn.Module):
def __init__(self, block, layers, strides=[1, 2, 2, 2], kernel3=[0, 0, 0, 0], num_classes=1000, avg_pool=True):
self.inplanes = 64
super(BagNet, self).__init__()
self.conv1 = nn.Conv2d(3, 64, kernel_size=1, stride=1, padding=0,
bias=False)
self.conv2 = nn.Conv2d(64, 64, kernel_size=3, stride=1, padding=0,
bias=False)
self.bn1 = nn.BatchNorm2d(64, momentum=0.001)
self.relu = nn.ReLU(inplace=True)
self.layer1 = self._make_layer(block, 64, layers[0], stride=strides[0], kernel3=kernel3[0], prefix='layer1')
self.layer2 = self._make_layer(block, 128, layers[1], stride=strides[1], kernel3=kernel3[1], prefix='layer2')
self.layer3 = self._make_layer(block, 256, layers[2], stride=strides[2], kernel3=kernel3[2], prefix='layer3')
self.layer4 = self._make_layer(block, 512, layers[3], stride=strides[3], kernel3=kernel3[3], prefix='layer4')
self.avgpool = nn.AvgPool2d(1, stride=1)
self.fc = nn.Linear(512 * block.expansion, num_classes)
self.avg_pool = avg_pool
self.block = block
for m in self.modules():
if isinstance(m, nn.Conv2d):
n = m.kernel_size[0] * m.kernel_size[1] * m.out_channels
m.weight.data.normal_(0, math.sqrt(2. / n))
elif isinstance(m, nn.BatchNorm2d):
m.weight.data.fill_(1)
m.bias.data.zero_()
def _make_layer(self, block, planes, blocks, stride=1, kernel3=0, prefix=''):
downsample = None
if stride != 1 or self.inplanes != planes * block.expansion:
downsample = nn.Sequential(
nn.Conv2d(self.inplanes, planes * block.expansion,
kernel_size=1, stride=stride, bias=False),
nn.BatchNorm2d(planes * block.expansion),
)
layers = []
kernel = 1 if kernel3 == 0 else 3
layers.append(block(self.inplanes, planes, stride, downsample, kernel_size=kernel))
self.inplanes = planes * block.expansion
for i in range(1, blocks):
kernel = 1 if kernel3 <= i else 3
layers.append(block(self.inplanes, planes, kernel_size=kernel))
return nn.Sequential(*layers)
def forward(self, x):
x = self.conv1(x)
x = self.conv2(x)
x = self.bn1(x)
x = self.relu(x)
x = self.layer1(x)
x = self.layer2(x)
x = self.layer3(x)
x = self.layer4(x)
if self.avg_pool:
x = nn.AvgPool2d(x.size()[2], stride=1)(x)
x = x.view(x.size(0), -1)
x = self.fc(x)
else:
x = x.permute(0,2,3,1)
x = self.fc(x)
return x
def bagnet33(pretrained=False, strides=[2, 2, 2, 1], **kwargs):
"""Constructs a Bagnet-33 model.
Args:
pretrained (bool): If True, returns a model pre-trained on ImageNet
"""
model = BagNet(Bottleneck, [3, 4, 6, 3], strides=strides, kernel3=[1,1,1,1], **kwargs)
if pretrained:
model.load_state_dict(model_zoo.load_url(model_urls['bagnet33']))
return model
def bagnet17(pretrained=False, strides=[2, 2, 2, 1], **kwargs):
"""Constructs a Bagnet-17 model.
Args:
pretrained (bool): If True, returns a model pre-trained on ImageNet
"""
model = BagNet(Bottleneck, [3, 4, 6, 3], strides=strides, kernel3=[1,1,1,0], **kwargs)
if pretrained:
model.load_state_dict(model_zoo.load_url(model_urls['bagnet17']))
return model
def bagnet9(pretrained=False, strides=[2, 2, 2, 1], **kwargs):
"""Constructs a Bagnet-9 model.
Args:
pretrained (bool): If True, returns a model pre-trained on ImageNet
"""
model = BagNet(Bottleneck, [3, 4, 6, 3], strides=strides, kernel3=[1,1,0,0], **kwargs)
if pretrained:
model.load_state_dict(model_zoo.load_url(model_urls['bagnet9']))
return model
# --- DeepGaze Adaptation ----
class RGBBagNet17(nn.Sequential):
def __init__(self):
super(RGBBagNet17, self).__init__()
self.bagnet = bagnet17(pretrained=True, avg_pool=False)
self.normalizer = Normalizer()
super(RGBBagNet17, self).__init__(self.normalizer, self.bagnet)
class RGBBagNet33(nn.Sequential):
def __init__(self):
super(RGBBagNet33, self).__init__()
self.bagnet = bagnet33(pretrained=True, avg_pool=False)
self.normalizer = Normalizer()
super(RGBBagNet33, self).__init__(self.normalizer, self.bagnet)
|