def __init__(self, stride=32, *args, **kwargs):
super(Resnet101, self).__init__()
assert stride in (8, 16, 32)
dils = [1, 1
] if stride == 32 else [el * (16 // stride) for el in (1, 2)]
strds = [2 if el == 1 else 1 for el in dils]
self.conv1 = nn.Conv2d(3,
64,
kernel_size=7,
stride=2,
padding=3,
bias=False)
self.bn1 = BatchNorm2d(64)
self.maxpool = nn.MaxPool2d(kernel_size=3,
stride=2,
padding=1,
dilation=1,
ceil_mode=False)
self.layer1 = create_stage(64, 256, 3, stride=1, dilation=1)
self.layer2 = create_stage(256, 512, 4, stride=2, dilation=1)
self.layer3 = create_stage(512,
1024,
23,
stride=strds[0],
dilation=dils[0])
self.layer4 = create_stage(1024,
2048,
3,
stride=strds[1],
dilation=dils[1])
self.init_weight()
def __init__(self,
in_chan,
out_chan,
stride=1,
stride_at_1x1=False,
dilation=1,
*args,
**kwargs):
super(Bottleneck, self).__init__(*args, **kwargs)
stride1x1, stride3x3 = (stride, 1) if stride_at_1x1 else (1, stride)
assert out_chan % 4 == 0
mid_chan = int(out_chan / 4)
self.conv1 = nn.Conv2d(in_chan,
mid_chan,
kernel_size=1,
stride=stride1x1,
bias=False)
self.bn1 = BatchNorm2d(mid_chan)
self.conv2 = nn.Conv2d(mid_chan,
mid_chan,
kernel_size=3,
stride=stride3x3,
padding=dilation,
dilation=dilation,
bias=False)
self.bn2 = BatchNorm2d(mid_chan)
self.conv3 = nn.Conv2d(mid_chan, out_chan, kernel_size=1, bias=False)
self.bn3 = BatchNorm2d(out_chan, activation='none')
self.relu = nn.ReLU(inplace=True)
self.downsample = None
if in_chan != out_chan or stride != 1:
self.downsample = nn.Sequential(
nn.Conv2d(in_chan,
out_chan,
kernel_size=1,
stride=stride,
bias=False), BatchNorm2d(out_chan,
activation='none'))
self.init_weight()
def __init__(self, in_chan, out_chan, ks=3, stride=1, padding=1, dilation=1, *args, **kwargs):
super(ConvBNReLU, self).__init__()
self.conv = nn.Conv2d(in_chan,
out_chan,
kernel_size = ks,
stride = stride,
padding = padding,
dilation = dilation,
bias = True)
self.bn = BatchNorm2d(out_chan)
self.init_weight()
def __init__(self,
in_chan,
out_chan,
ks=3,
stride=1,
pad=0,
dilation=1,
slope=0.1,
*args,
**kwargs):
super(ConvBNLeaky, self).__init__()
self.conv = nn.Conv2d(in_chan,
out_chan,
kernel_size=ks,
padding=pad,
dilation=dilation,
stride=stride,
bias=False)
self.bn = BatchNorm2d(out_chan, slope=slope)
self._init_weight()
def __init__(self,
in_chan,
out_chan,
ks=3,
stride=1,
padding=1,
dilation=1,
bias=False,
*args,
**kwargs):
super(SepConvBNReLU, self).__init__()
self.conv = nn.Conv2d(in_chan,
in_chan,
kernel_size=ks,
stride=stride,
padding=dilation,
dilation=dilation,
groups=in_chan,
bias=bias)
self.bn = BatchNorm2d(in_chan)
self.pairwise = nn.Conv2d(in_chan, out_chan, kernel_size=1, bias=bias)
self.init_weight()