deepdraw.models.make_layers#
Functions
|
|
|
|
|
|
|
|
Classes
|
|
|
|
|
Combines Conv2d, ConvTransposed2d and Cropping. |
- deepdraw.models.make_layers.conv_with_kaiming_uniform(in_channels, out_channels, kernel_size, stride=1, padding=0, dilation=1)[source]#
- deepdraw.models.make_layers.convtrans_with_kaiming_uniform(in_channels, out_channels, kernel_size, stride=1, padding=0, dilation=1)[source]#
- class deepdraw.models.make_layers.UpsampleCropBlock(in_channels, out_channels, up_kernel_size, up_stride, up_padding, pixelshuffle=False)[source]#
Bases:
Module
Combines Conv2d, ConvTransposed2d and Cropping. Simulates the caffe2 crop layer in the forward function.
Used for DRIU and HED.
- Parameters:
- deepdraw.models.make_layers.icnr(x, scale=2, init=<function kaiming_normal_>)[source]#
https://docs.fast.ai/layers.html#PixelShuffle_ICNR.
ICNR init of
x
, withscale
andinit
function.
- class deepdraw.models.make_layers.PixelShuffle_ICNR(ni, nf=None, scale=2)[source]#
Bases:
Module
https://docs.fast.ai/layers.html#PixelShuffle_ICNR.
Upsample by
scale
fromni
filters tonf
(defaultni
), usingtorch.nn.PixelShuffle
,icnr
init, andweight_norm
.- forward(x)[source]#
Defines the computation performed at every call.
Should be overridden by all subclasses.
Note
Although the recipe for forward pass needs to be defined within this function, one should call the
Module
instance afterwards instead of this since the former takes care of running the registered hooks while the latter silently ignores them.
- class deepdraw.models.make_layers.UnetBlock(up_in_c, x_in_c, pixel_shuffle=False, middle_block=False)[source]#
Bases:
Module
- forward(up_in, x_in)[source]#
Defines the computation performed at every call.
Should be overridden by all subclasses.
Note
Although the recipe for forward pass needs to be defined within this function, one should call the
Module
instance afterwards instead of this since the former takes care of running the registered hooks while the latter silently ignores them.