Rename normalize to norm_cfg (#637)

* rename normalize to norm_cfg

* update configs

* Update resnet.py

mmdet/models/necks/fpn.py

@@ -18,7 +18,7 @@ class FPN(nn.Module):
                  add_extra_convs=False,
                  extra_convs_on_inputs=True,
                  conv_cfg=None,
-                 normalize=None,
+                 norm_cfg=None,
                  activation=None):
         super(FPN, self).__init__()
         assert isinstance(in_channels, list)
@@ -27,7 +27,6 @@ class FPN(nn.Module):
         self.num_ins = len(in_channels)
         self.num_outs = num_outs
         self.activation = activation
-        self.with_bias = normalize is None
 
         if end_level == -1:
             self.backbone_end_level = self.num_ins
@@ -51,8 +50,7 @@ class FPN(nn.Module):
                 out_channels,
                 1,
                 conv_cfg=conv_cfg,
-                normalize=normalize,
-                bias=self.with_bias,
+                norm_cfg=norm_cfg,
                 activation=self.activation,
                 inplace=False)
             fpn_conv = ConvModule(
@@ -61,8 +59,7 @@ class FPN(nn.Module):
                 3,
                 padding=1,
                 conv_cfg=conv_cfg,
-                normalize=normalize,
-                bias=self.with_bias,
+                norm_cfg=norm_cfg,
                 activation=self.activation,
                 inplace=False)
 
@@ -83,8 +80,8 @@ class FPN(nn.Module):
                     3,
                     stride=2,
                     padding=1,
-                    normalize=normalize,
-                    bias=self.with_bias,
+                    conv_cfg=conv_cfg,
+                    norm_cfg=norm_cfg,
                     activation=self.activation,
                     inplace=False)
                 self.fpn_convs.append(extra_fpn_conv)

mmdet/models/shared_heads/res_layer.py

@@ -17,13 +17,13 @@ class ResLayer(nn.Module):
                  stride=2,
                  dilation=1,
                  style='pytorch',
-                 normalize=dict(type='BN', requires_grad=True),
+                 norm_cfg=dict(type='BN', requires_grad=True),
                  norm_eval=True,
                  with_cp=False,
                  dcn=None):
         super(ResLayer, self).__init__()
         self.norm_eval = norm_eval
-        self.normalize = normalize
+        self.norm_cfg = norm_cfg
         self.stage = stage
         block, stage_blocks = ResNet.arch_settings[depth]
         stage_block = stage_blocks[stage]
@@ -39,7 +39,7 @@ class ResLayer(nn.Module):
             dilation=dilation,
             style=style,
             with_cp=with_cp,
-            normalize=self.normalize,
+            norm_cfg=self.norm_cfg,
             dcn=dcn)
         self.add_module('layer{}'.format(stage + 1), res_layer)
 

mmdet/models/utils/conv_module.py

@@ -42,6 +42,27 @@ def build_conv_layer(cfg, *args, **kwargs):
 
 
 class ConvModule(nn.Module):
+    """Conv-Norm-Activation block.
+
+    Args:
+        in_channels (int): Same as nn.Conv2d.
+        out_channels (int): Same as nn.Conv2d.
+        kernel_size (int or tuple[int]): Same as nn.Conv2d.
+        stride (int or tuple[int]): Same as nn.Conv2d.
+        padding (int or tuple[int]): Same as nn.Conv2d.
+        dilation (int or tuple[int]): Same as nn.Conv2d.
+        groups (int): Same as nn.Conv2d.
+        bias (bool or str): If specified as `auto`, it will be decided by the
+            norm_cfg. Bias will be set as True if norm_cfg is None, otherwise
+            False.
+        conv_cfg (dict): Config dict for convolution layer.
+        norm_cfg (dict): Config dict for normalization layer.
+        activation (str or None): Activation type, "ReLU" by default.
+        inplace (bool): Whether to use inplace mode for activation.
+        activate_last (bool): Whether to apply the activation layer in the
+            last. (Do not use this flag since the behavior and api may be
+            changed in the future.)
+    """
 
     def __init__(self,
                  in_channels,
@@ -51,35 +72,42 @@ class ConvModule(nn.Module):
                  padding=0,
                  dilation=1,
                  groups=1,
-                 bias=True,
+                 bias='auto',
                  conv_cfg=None,
-                 normalize=None,
+                 norm_cfg=None,
                  activation='relu',
                  inplace=True,
                  activate_last=True):
         super(ConvModule, self).__init__()
         assert conv_cfg is None or isinstance(conv_cfg, dict)
-        assert normalize is None or isinstance(normalize, dict)
-        self.with_norm = normalize is not None
-        self.with_activatation = activation is not None
-        self.with_bias = bias
+        assert norm_cfg is None or isinstance(norm_cfg, dict)
+        self.conv_cfg = conv_cfg
+        self.norm_cfg = norm_cfg
         self.activation = activation
+        self.inplace = inplace
         self.activate_last = activate_last
 
+        self.with_norm = norm_cfg is not None
+        self.with_activatation = activation is not None
+        # if the conv layer is before a norm layer, bias is unnecessary.
+        if bias == 'auto':
+            bias = False if self.with_norm else True
+        self.with_bias = bias
+
         if self.with_norm and self.with_bias:
             warnings.warn('ConvModule has norm and bias at the same time')
 
-        self.conv = build_conv_layer(
-            conv_cfg,
+        # build convolution layer
+        self.conv = build_conv_layer(conv_cfg,
                                      in_channels,
                                      out_channels,
                                      kernel_size,
-            stride,
-            padding,
-            dilation,
-            groups,
+                                     stride=stride,
+                                     padding=padding,
+                                     dilation=dilation,
+                                     groups=groups,
                                      bias=bias)
-
+        # export the attributes of self.conv to a higher level for convenience
         self.in_channels = self.conv.in_channels
         self.out_channels = self.conv.out_channels
         self.kernel_size = self.conv.kernel_size
@@ -90,17 +118,21 @@ class ConvModule(nn.Module):
         self.output_padding = self.conv.output_padding
         self.groups = self.conv.groups
 
+        # build normalization layers
         if self.with_norm:
             norm_channels = out_channels if self.activate_last else in_channels
-            self.norm_name, norm = build_norm_layer(normalize, norm_channels)
+            self.norm_name, norm = build_norm_layer(norm_cfg, norm_channels)
             self.add_module(self.norm_name, norm)
 
+        # build activation layer
         if self.with_activatation:
-            assert activation in ['relu'], 'Only ReLU supported.'
+            if self.activation not in ['relu']:
+                raise ValueError('{} is currently not supported.'.format(
+                    self.activation))
             if self.activation == 'relu':
                 self.activate = nn.ReLU(inplace=inplace)
 
-        # Default using msra init
+        # Use msra init by default
         self.init_weights()
 
     @property
@@ -121,6 +153,7 @@ class ConvModule(nn.Module):
             if activate and self.with_activatation:
                 x = self.activate(x)
         else:
+            # WARN: this may be removed or modified
             if norm and self.with_norm:
                 x = self.norm(x)
             if activate and self.with_activatation: