mirrors
/
mmsegmentation
mirror of https://github.com/open-mmlab/mmsegmentation.git
1
0
Fork 0
Code Issues Projects Releases Wiki Activity

Docstrings fixed.

pull/58/head
johnzja 2020-08-12 15:36:45 +08:00
parent 568106c2e2
commit a3fb9b1f51
1 changed files with 14 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

23
mmseg/models/backbones/fast_scnn.py
View File

@ -73,15 +73,20 @@ class GlobalFeatureExtractor(nn.Module):
Args: Args:
in_channels (int): Number of input channels of the GFE module. in_channels (int): Number of input channels of the GFE module.
block_channels (tuple): Tuple of ints. Each int specifies the Default: 64
block_channels (tuple[int]): Tuple of ints. Each int specifies the
number of output channels of each Inverted Residual module. number of output channels of each Inverted Residual module.
Default: (64, 96, 128)
out_channels(int): Number of output channels of the GFE module. out_channels(int): Number of output channels of the GFE module.
expand_ratio (int): upsampling factor of each Inverted Residual Default: 128
module. expand_ratio (int): Upsampling factor of each Inverted Residual
num_blocks (tuple): Tuple of ints. Each int specifies the number of module. Default: 6
times each Inverted Residual module is repeated. num_blocks (tuple[int]): Tuple of ints. Each int specifies the
pool_scales (tuple): Tuple of ints. Each int specifies the parameter number of times each Inverted Residual module is repeated.
required in 'global average pooling' within PPM. Default: (3, 3, 3)
pool_scales (tuple[int]): Tuple of ints. Each int specifies
the parameter required in 'global average pooling' within PPM.
Default: (1, 2, 3, 6)
conv_cfg (dict | None): Config of conv layers. Default: None conv_cfg (dict | None): Config of conv layers. Default: None
norm_cfg (dict | None): Config of norm layers. Default: norm_cfg (dict | None): Config of norm layers. Default:
dict(type='BN') dict(type='BN')
@ -229,7 +234,7 @@ class FastSCNN(nn.Module):
"""Fast-SCNN Backbone. """Fast-SCNN Backbone.
Args: Args:
in_channels (int): Number of input image channels. Default: 3. in_channels (int): Number of input image channels. Default: 3.
downsample_dw_channels (tuple): Number of output channels after downsample_dw_channels (tuple[int]): Number of output channels after
the first conv layer & the second conv layer in the first conv layer & the second conv layer in
Learning-To-Downsample (LTD) module. Learning-To-Downsample (LTD) module.
Default: (32, 48). Default: (32, 48).
@ -237,7 +242,7 @@ class FastSCNN(nn.Module):
Global Feature Extractor(GFE). Global Feature Extractor(GFE).
Equal to number of output channels of LTD. Equal to number of output channels of LTD.
Default: 64. Default: 64.
global_block_channels (tuple): Tuple of integers that describe global_block_channels (tuple[int]): Tuple of integers that describe
the output channels for each of the MobileNet-v2 bottleneck the output channels for each of the MobileNet-v2 bottleneck
residual blocks in GFE. residual blocks in GFE.
Default: (64, 96, 128). Default: (64, 96, 128).