def __init__(self,
in_channels,
inner_channels=128,
fpem_repeat=2,
**kwargs):
"""
PANnet
:param in_channels: 基础网络输出的维度
"""
super().__init__()
self.conv_out = inner_channels
inplace = True
# reduce layers
self.reduce_conv_c2 = ConvBnRelu(in_channels[0],
inner_channels,
kernel_size=1,
inplace=inplace)
self.reduce_conv_c3 = ConvBnRelu(in_channels[1],
inner_channels,
kernel_size=1,
inplace=inplace)
self.reduce_conv_c4 = ConvBnRelu(in_channels[2],
inner_channels,
kernel_size=1,
inplace=inplace)
self.reduce_conv_c5 = ConvBnRelu(in_channels[3],
inner_channels,
kernel_size=1,
inplace=inplace)
self.fpems = nn.ModuleList()
for i in range(fpem_repeat):
self.fpems.append(FPEM(self.conv_out))
self.out_channels = self.conv_out * 4
def __init__(self, in_channels, inner_channels=256, **kwargs):
"""
:param in_channels: 基础网络输出的维度
:param kwargs:
"""
super().__init__()
inplace = True
self.conv_out = inner_channels
inner_channels = inner_channels // 4
# reduce layers
self.reduce_conv_c2 = ConvBnRelu(in_channels[0], inner_channels, kernel_size=1, inplace=inplace)
self.reduce_conv_c3 = ConvBnRelu(in_channels[1], inner_channels, kernel_size=1, inplace=inplace)
self.reduce_conv_c4 = ConvBnRelu(in_channels[2], inner_channels, kernel_size=1, inplace=inplace)
self.reduce_conv_c5 = ConvBnRelu(in_channels[3], inner_channels, kernel_size=1, inplace=inplace)
# Smooth layers
self.smooth_p4 = ConvBnRelu(inner_channels, inner_channels, kernel_size=3, padding=1, inplace=inplace)
self.smooth_p3 = ConvBnRelu(inner_channels, inner_channels, kernel_size=3, padding=1, inplace=inplace)
self.smooth_p2 = ConvBnRelu(inner_channels, inner_channels, kernel_size=3, padding=1, inplace=inplace)
self.conv = nn.Sequential(
nn.Conv2d(self.conv_out, self.conv_out, kernel_size=3, padding=1, stride=1),
nn.BatchNorm2d(self.conv_out),
nn.ReLU(inplace=inplace)
)
self.out_channels = self.conv_out
def __init__(self, in_channels, inner_channels=256, **kwargs):
"""
:param in_channels: 基础网络输出的维度 [64, 128, 256, 512]
:param kwargs:
"""
super().__init__()
inplace = True
self.conv_out = inner_channels
inner_channels = inner_channels // 4 # 256 // 4 = 64
# reduce layers
self.reduce_conv_c2 = ConvBnRelu(in_channels[0],
inner_channels,
kernel_size=1,
inplace=inplace)
self.reduce_conv_c3 = ConvBnRelu(in_channels[1],
inner_channels,
kernel_size=1,
inplace=inplace)
self.reduce_conv_c4 = ConvBnRelu(in_channels[2],
inner_channels,
kernel_size=1,
inplace=inplace)
self.reduce_conv_c5 = ConvBnRelu(in_channels[3],
inner_channels,
kernel_size=1,
inplace=inplace)
# Smooth layers
self.smooth_p4 = ConvBnRelu(inner_channels,
inner_channels,
kernel_size=3,
padding=1,
inplace=inplace) # 311
self.smooth_p3 = ConvBnRelu(inner_channels,
inner_channels,
kernel_size=3,
padding=1,
inplace=inplace)
self.smooth_p2 = ConvBnRelu(inner_channels,
inner_channels,
kernel_size=3,
padding=1,
inplace=inplace)
#self.upsample = nn.Upsample(scale_factor=2, mode='nearest')
self.conv = nn.Sequential(
nn.Conv2d(self.conv_out,
self.conv_out,
kernel_size=3,
padding=1,
stride=1), nn.BatchNorm2d(self.conv_out),
nn.ReLU(inplace=inplace))
self.out_channels = self.conv_out
def __init__(self,
model_config: dict,
layers=[2, 2, 2, 2],
in_channels=3,
inner_channels=256,
k=50):
"""
PANnet
:param model_config: 模型配置
"""
super().__init__()
self.name = f'resnet18_fpn_db'
self.inplanes = 64
self.backon_out_channels = []
self.conv1 = nn.Conv2d(in_channels,
64,
kernel_size=7,
stride=2,
padding=3,
bias=False)
self.bn1 = nn.BatchNorm2d(64)
self.relu = nn.ReLU(inplace=True)
self.maxpool = nn.MaxPool2d(kernel_size=3, stride=2, padding=1)
self.layer1 = self._make_layer(BasicBlock, 64, layers[0])
self.layer2 = self._make_layer(BasicBlock, 128, layers[1], stride=2)
self.layer3 = self._make_layer(BasicBlock, 256, layers[2], stride=2)
self.layer4 = self._make_layer(BasicBlock, 512, layers[3], stride=2)
inplace = True
self.conv_out = inner_channels
inner_channels = inner_channels // 4
# reduce layers
self.reduce_conv_c2 = ConvBnRelu(self.backon_out_channels[0],
inner_channels,
kernel_size=1,
inplace=inplace)
self.reduce_conv_c3 = ConvBnRelu(self.backon_out_channels[1],
inner_channels,
kernel_size=1,
inplace=inplace)
self.reduce_conv_c4 = ConvBnRelu(self.backon_out_channels[2],
inner_channels,
kernel_size=1,
inplace=inplace)
self.reduce_conv_c5 = ConvBnRelu(self.backon_out_channels[3],
inner_channels,
kernel_size=1,
inplace=inplace)
# Smooth layers
self.smooth_p4 = ConvBnRelu(inner_channels,
inner_channels,
kernel_size=3,
padding=1,
inplace=inplace)
self.smooth_p3 = ConvBnRelu(inner_channels,
inner_channels,
kernel_size=3,
padding=1,
inplace=inplace)
self.smooth_p2 = ConvBnRelu(inner_channels,
inner_channels,
kernel_size=3,
padding=1,
inplace=inplace)
self.conv = nn.Sequential(
nn.Conv2d(self.conv_out,
self.conv_out,
kernel_size=3,
padding=1,
stride=1), nn.BatchNorm2d(self.conv_out),
nn.ReLU(inplace=inplace))
self.out_channels = self.conv_out
self.k = k
self.binarize = nn.Sequential(
nn.Conv2d(self.out_channels, self.out_channels // 4, 3, padding=1),
nn.BatchNorm2d(self.out_channels // 4), nn.ReLU(inplace=True),
nn.ConvTranspose2d(self.out_channels // 4, self.out_channels // 4,
2, 2), nn.BatchNorm2d(self.out_channels // 4),
nn.ReLU(inplace=True),
nn.ConvTranspose2d(self.out_channels // 4, 1, 2, 2), nn.Sigmoid())
self.binarize.apply(self.weights_init)
self.thresh = self._init_thresh(self.out_channels)
self.thresh.apply(self.weights_init)