def __init__(self, inchannel, outchannel, stride=1, x=0):
super(ResidualBlock, self).__init__()
self.x = x
self.left = nn.Sequential(
nn.Conv2d(inchannel,
outchannel,
kernel_size=3,
stride=stride,
padding=1,
bias=False), nn.BatchNorm2d(outchannel),
nn.ReLU(inplace=True),
nn.Conv2d(outchannel,
outchannel,
kernel_size=3,
stride=1,
padding=1,
bias=False), nn.BatchNorm2d(outchannel))
self.x = cal_conv2d(x=self.x, kernel_size=3, stride=stride, padding=1)
self.x = cal_conv2d(x=self.x, kernel_size=3, stride=1, padding=1)
self.shortcut = nn.Sequential()
if stride != 1 or inchannel != outchannel:
self.shortcut = nn.Sequential(
nn.Conv2d(inchannel,
outchannel,
kernel_size=1,
stride=stride,
bias=False), nn.BatchNorm2d(outchannel))
self.x = cal_conv2d(x=self.x, kernel_size=1, stride=stride)
def __init__(self, ResidualBlock, x=32, num_classes=10):
super(ResNet, self).__init__()
self.inchannel = 64
self.INPUT_SIZE = x
self.conv1 = nn.Sequential(
nn.Conv2d(3, 64, kernel_size=3, stride=1, padding=1, bias=False),
nn.BatchNorm2d(64),
nn.ReLU(),
)
self.INPUT_SIZE = cal_conv2d(x=self.INPUT_SIZE,
kernel_size=3,
stride=1,
padding=1)
self.layer1 = self.make_layer(ResidualBlock, 64, 2, stride=1)
self.layer2 = self.make_layer(ResidualBlock, 128, 2, stride=2)
self.layer3 = self.make_layer(ResidualBlock, 256, 2, stride=2)
self.layer4 = self.make_layer(ResidualBlock, 512, 2, stride=2)
print(self.INPUT_SIZE)
self.fc = nn.Linear(512 * self.INPUT_SIZE * self.INPUT_SIZE,
num_classes)
def __init__(self,
ResidualBlock,
cfg,
x=32,
num_classes=10,
ipt_channel=3):
super(ResNet, self).__init__()
self.inchannel = 16
self.INPUT_SIZE = x
self.conv1 = nn.Sequential(
nn.Conv2d(ipt_channel,
16,
kernel_size=2,
stride=1,
padding=1,
bias=False),
nn.BatchNorm2d(16),
nn.ReLU(),
)
self.INPUT_SIZE = cal_conv2d(x=self.INPUT_SIZE,
kernel_size=2,
stride=1,
padding=1)
layer = []
self.cfg_length = len(cfg)
for i in range(self.cfg_length):
layer.append(
self.make_layer(ResidualBlock, cfg[i]['channels'],
cfg[i]['num_blocks'], cfg[i]['stride']))
if i == self.cfg_length - 1:
# print(self.INPUT_SIZE)
self.INPUT_SIZE = cal_maxpool2d(x=self.INPUT_SIZE,
kernel_size=2)
# print('I am')
# print(self.INPUT_SIZE)
self.layers = nn.Sequential(*layer)
self.pool = nn.MaxPool2d(kernel_size=2)
self.fc = nn.Linear(LINEAR_OUTPUT, num_classes)