fastreid.modeling.backbones.osnet.OSBlock Class Reference

Inheritance diagram for fastreid.modeling.backbones.osnet.OSBlock:

Public Member Functions
	__init__ (self, in_channels, out_channels, bn_norm, IN=False, bottleneck_reduction=4, **kwargs)
	forward (self, x)

Public Attributes
	conv1
	conv2a
	conv2b
	conv2c
	conv2d
	gate
	conv3
	downsample
	IN
	relu

Detailed Description

Omni-scale feature learning block.

Definition at line 229 of file osnet.py.

Constructor & Destructor Documentation

__init__()

fastreid.modeling.backbones.osnet.OSBlock.__init__	(		self,
			in_channels,
			out_channels,
			bn_norm,
			IN = False,
			bottleneck_reduction = 4,
		**	kwargs )

Definition at line 232 of file osnet.py.

    ):
        super(OSBlock, self).__init__()
        mid_channels = out_channels // bottleneck_reduction
        self.conv1 = Conv1x1(in_channels, mid_channels, bn_norm)
        self.conv2a = LightConv3x3(mid_channels, mid_channels, bn_norm)
        self.conv2b = nn.Sequential(
            LightConv3x3(mid_channels, mid_channels, bn_norm),
            LightConv3x3(mid_channels, mid_channels, bn_norm),
        )
        self.conv2c = nn.Sequential(
            LightConv3x3(mid_channels, mid_channels, bn_norm),
            LightConv3x3(mid_channels, mid_channels, bn_norm),
            LightConv3x3(mid_channels, mid_channels, bn_norm),
        )
        self.conv2d = nn.Sequential(
            LightConv3x3(mid_channels, mid_channels, bn_norm),
            LightConv3x3(mid_channels, mid_channels, bn_norm),
            LightConv3x3(mid_channels, mid_channels, bn_norm),
            LightConv3x3(mid_channels, mid_channels, bn_norm),
        )
        self.gate = ChannelGate(mid_channels)
        self.conv3 = Conv1x1Linear(mid_channels, out_channels, bn_norm)
        self.downsample = None
        if in_channels != out_channels:
            self.downsample = Conv1x1Linear(in_channels, out_channels, bn_norm)
        self.IN = None
        if IN: self.IN = nn.InstanceNorm2d(out_channels, affine=True)
        self.relu = nn.ReLU(True)
 

Member Function Documentation

forward()

fastreid.modeling.backbones.osnet.OSBlock.forward	(		self,
			x )

Definition at line 269 of file osnet.py.

    def forward(self, x):
        identity = x
        x1 = self.conv1(x)
        x2a = self.conv2a(x1)
        x2b = self.conv2b(x1)
        x2c = self.conv2c(x1)
        x2d = self.conv2d(x1)
        x2 = self.gate(x2a) + self.gate(x2b) + self.gate(x2c) + self.gate(x2d)
        x3 = self.conv3(x2)
        if self.downsample is not None:
            identity = self.downsample(identity)
        out = x3 + identity
        if self.IN is not None:
            out = self.IN(out)
        return self.relu(out)
 
 

Member Data Documentation

conv1

fastreid.modeling.backbones.osnet.OSBlock.conv1

Definition at line 243 of file osnet.py.

conv2a

fastreid.modeling.backbones.osnet.OSBlock.conv2a

Definition at line 244 of file osnet.py.

conv2b

fastreid.modeling.backbones.osnet.OSBlock.conv2b

Definition at line 245 of file osnet.py.

conv2c

fastreid.modeling.backbones.osnet.OSBlock.conv2c

Definition at line 249 of file osnet.py.

conv2d

fastreid.modeling.backbones.osnet.OSBlock.conv2d

Definition at line 254 of file osnet.py.

conv3

fastreid.modeling.backbones.osnet.OSBlock.conv3

Definition at line 261 of file osnet.py.

downsample

fastreid.modeling.backbones.osnet.OSBlock.downsample

Definition at line 262 of file osnet.py.

gate

fastreid.modeling.backbones.osnet.OSBlock.gate

Definition at line 260 of file osnet.py.

IN

fastreid.modeling.backbones.osnet.OSBlock.IN

Definition at line 265 of file osnet.py.

relu

fastreid.modeling.backbones.osnet.OSBlock.relu

Definition at line 267 of file osnet.py.

---

The documentation for this class was generated from the following file:

• smreid/fastreid/modeling/backbones/osnet.py