conv_bn

plugins.extract.detect.retinaface.conv_bn(in_channels: int, out_channels: int, kernel: int = 3, stride: int = 1, padding: int = 1, use_relu: bool = False, leaky: float = 0.0) → Sequential

Generates a Conv Batch Norm sequential module for RetinaFace

Parameters:

  • in_channels (int) – The number of input channels

  • out_channels (int) – The number of output channels

  • kernel (int) – The kernel size. Default: 3

  • stride (int) – The number of strides. Default: 1

  • padding (int) – The padding to apply. Default: 1

  • use_relu (bool) – True to use LeakyReLU activation

  • leaky (float) – The negative float value for the LeakyReLU. Default: 0.0

Return type:

The built sequential module