本文整理匯總了Python中mmcv.cnn.VGG屬性的典型用法代碼示例。如果您正苦於以下問題：Python cnn.VGG屬性的具體用法？Python cnn.VGG怎麽用？Python cnn.VGG使用的例子？那麽, 這裏精選的屬性代碼示例或許可以為您提供幫助。您也可以進一步了解該屬性所在類mmcv.cnn的用法示例。
在下文中一共展示了cnn.VGG屬性的1個代碼示例，這些例子默認根據受歡迎程度排序。您可以為喜歡或者感覺有用的代碼點讚，您的評價將有助於係統推薦出更棒的Python代碼示例。
示例1: __init__
# 需要導入模塊: from mmcv import cnn [as 別名]
# 或者: from mmcv.cnn import VGG [as 別名]
def __init__(self,
input_size,
depth,
with_last_pool=False,
ceil_mode=True,
out_indices=(3, 4),
out_feature_indices=(22, 34),
l2_norm_scale=20.):
# TODO: in_channels for mmcv.VGG
super(SSDVGG, self).__init__(
depth,
with_last_pool=with_last_pool,
ceil_mode=ceil_mode,
out_indices=out_indices)
assert input_size in (300, 512)
self.input_size = input_size
self.features.add_module(
str(len(self.features)),
nn.MaxPool2d(kernel_size=3, stride=1, padding=1))
self.features.add_module(
str(len(self.features)),
nn.Conv2d(512, 1024, kernel_size=3, padding=6, dilation=6))
self.features.add_module(
str(len(self.features)), nn.ReLU(inplace=True))
self.features.add_module(
str(len(self.features)), nn.Conv2d(1024, 1024, kernel_size=1))
self.features.add_module(
str(len(self.features)), nn.ReLU(inplace=True))
self.out_feature_indices = out_feature_indices
self.inplanes = 1024
self.extra = self._make_extra_layers(self.extra_setting[input_size])
self.l2_norm = L2Norm(
self.features[out_feature_indices[0] - 1].out_channels,
l2_norm_scale)