def __init__(self, n_classes, n_blocks):
super(ResNet8s, self).__init__()
self.add_module(
'layer1',
nn.Sequential(
OrderedDict([
('conv1', _ConvBatchNormReLU(3, 64, 7, 2, 3, 1)),
('pool', nn.MaxPool2d(3, 2, 1, ceil_mode=True)),
])
)
)
self.add_module('layer2', _ResBlock(n_blocks[0], 64, 64, 256, 1, 1))
self.add_module('layer3', _ResBlock(n_blocks[1], 256, 128, 512, 2, 1))
self.add_module('layer4', _ResBlock(n_blocks[2], 512, 256, 1024, 1, 2))
self.add_module('layer5', _ResBlock(n_blocks[3], 1024, 512, 2048, 1, 4))
self.classifier = nn.Sequential(
OrderedDict([
('conv5_4', _ConvBatchNormReLU(2048, 512, 3, 1, 1, 1)),
('drop5_4', nn.Dropout2d(p=0.1)),
('conv6', nn.Conv2d(512, n_classes, 1, stride=1, padding=0)),
])
)
def __init__(self, in_channels, pyramids=[6, 3, 2, 1]):
super(_PyramidPoolModule, self).__init__()
out_channels = in_channels // len(pyramids)
self.stages = nn.Module()
for i, p in enumerate(pyramids):
self.stages.add_module(
's{}'.format(i),
nn.Sequential(
OrderedDict([
('pool', nn.AdaptiveAvgPool2d(output_size=p)),
('conv', _ConvBatchNormReLU(in_channels, out_channels, 1, 1, 0, 1)),
])
)
)
def __init__(self, n_classes, n_blocks, pyramids, multi_grid=[1, 2, 1]):
super(DeepLabV3, self).__init__()
self.add_module(
'layer1',
nn.Sequential(
OrderedDict([
('conv1', _ConvBatchNormReLU(3, 64, 7, 2, 3, 1)),
('pool', nn.MaxPool2d(3, 2, 1, ceil_mode=True)),
])))
self.add_module('layer2', _ResBlock(n_blocks[0], 64, 64, 256, 1,
1)) # output_stride=4
self.add_module('layer3', _ResBlock(n_blocks[1], 256, 128, 512, 2,
1)) # output_stride=8
self.add_module('layer4', _ResBlock(n_blocks[2], 512, 256, 1024, 1,
2)) # output_stride=8
self.add_module(
'layer5',
_ResBlockMG(n_blocks[3], 1024, 512, 2048, 1, 2, mg=multi_grid))
self.add_module('aspp', _ASPPModule(2048, 256, pyramids))
self.add_module(
'fc1',
_ConvBatchNormReLU(256 * (len(pyramids) + 2), 256, 1, 1, 0, 1))
self.add_module('fc2', nn.Conv2d(256, n_classes, kernel_size=1))
def __init__(self, n_classes, n_blocks, pyramids):
super(DeepLabV2, self).__init__()
self.add_module(
'layer1',
nn.Sequential(
OrderedDict([('conv1', _ConvBatchNormReLU(3, 64, 7, 2, 3, 1)),
('pool', nn.MaxPool2d(3, 2, 1,
ceil_mode=True))])))
self.add_module('layer2', _ResBlock(n_blocks[0], 64, 64, 256, 1, 1))
self.add_module('layer3', _ResBlock(n_blocks[1], 256, 128, 512, 2, 1))
self.add_module('layer4', _ResBlock(n_blocks[2], 512, 256, 1024, 1, 2))
self.add_module('layer5', _ResBlock(n_blocks[3], 1024, 512, 2048, 1,
4)) # NOQA
self.add_module('aspp', _ASPPModule(2048, n_classes, pyramids))
self._freeze_bn()
def __init__(self, n_classes, n_blocks, pyramids):
super(DeepLabV2, self).__init__()
self.add_module(
'layer1',
nn.Sequential(
OrderedDict([
('conv1', _ConvBatchNormReLU(3, 64, 7, 2, 3, 1)),
('pool', nn.MaxPool2d(3, 2, 1, ceil_mode=True)),
])))
self.add_module('layer2', _ResBlock(n_blocks[0], 64, 64, 256, 1, 1))
self.add_module('layer3', _ResBlock(n_blocks[1], 256, 128, 512, 2, 1))
self.add_module('layer4', _ResBlock(n_blocks[2], 512, 256, 1024, 1, 2))
self.add_module('layer5', _ResBlock(n_blocks[3], 1024, 512, 2048, 1,
4))
# self.add_module('aspp', _ASPPModule(2048, n_classes, pyramids)) Arezoo
self.add_module('aspp', _ASPPModule(2048, 4, pyramids))
self.upsample_1 = nn.Sequential(
nn.Conv2d(4, 4, kernel_size=1, stride=1, padding=0),
nn.BatchNorm2d(4),
Interpolate(),
)
self.weight_init()