def ResNeXt50(include_top=True,
weights='imagenet',
input_tensor=None,
input_shape=None,
pooling=None,
classes=1000,
**kwargs):
"""Instantiates the ResNeXt50 architecture."""
def stack_fn(x):
x = stack1(x, 64, 3, stride1=1, groups=32, base_width=4, name='conv2')
x = stack1(x, 128, 4, groups=32, base_width=4, name='conv3')
x = stack1(x, 256, 6, groups=32, base_width=4, name='conv4')
return stack1(x, 512, 3, groups=32, base_width=4, name='conv5')
return ResNet(stack_fn, False, 'resnext50', include_top, weights,
input_tensor, input_shape, pooling, False, None, classes,
**kwargs)
def WideResNet101(include_top=True,
weights='imagenet',
input_tensor=None,
input_shape=None,
pooling=None,
classes=1000,
**kwargs):
"""Instantiates the Wide-ResNet101-2 architecture."""
def stack_fn(x):
x = stack1(x, 64, 3, stride1=1, base_width=128, name='conv2')
x = stack1(x, 128, 4, base_width=128, name='conv3')
x = stack1(x, 256, 23, base_width=128, name='conv4')
return stack1(x, 512, 3, base_width=128, name='conv5')
return ResNet(stack_fn, False, 'wide_resnet101', include_top, weights,
input_tensor, input_shape, pooling, False, None, classes,
**kwargs)
def ResNet34(include_top=True,
weights='imagenet',
input_tensor=None,
input_shape=None,
pooling=None,
classes=1000,
**kwargs):
"""Instantiates the ResNet34 architecture."""
def stack_fn(x):
x = stack3(x, 64, 3, stride1=1, conv_shortcut=False, name='conv2')
x = stack3(x, 128, 4, name='conv3')
x = stack3(x, 256, 6, name='conv4')
return stack3(x, 512, 3, name='conv5')
return ResNet(stack_fn, False, 'resnet34', include_top, weights,
input_tensor, input_shape, pooling, False, None, classes,
**kwargs)
def ResNeSt50(include_top=True,
weights='imagenet',
input_tensor=None,
input_shape=None,
pooling=None,
classes=1000,
**kwargs):
"""Instantiates the ResNeSt50 architecture."""
def stack_fn(x):
x = stack2(x,
64,
3,
stride1=1,
base_width=64,
radix=2,
is_first=False,
name='conv2')
x = stack2(x, 128, 4, base_width=64, radix=2, name='conv3')
x = stack2(x, 256, 6, base_width=64, radix=2, name='conv4')
return stack2(x, 512, 3, base_width=64, radix=2, name='conv5')
return ResNet(stack_fn, False, 'resnest50', include_top, weights,
input_tensor, input_shape, pooling, True, 32, classes,
**kwargs)