basecls.models.hrnet#
HRNet Series
HRNet: “Deep High-Resolution Representation Learning for Visual Recognition”
引用
HRNet/HRNet-Image-Classification
- class basecls.models.hrnet.UpsampleNearest(scale_factor)[源代码]#
基类:
Module
Nearest upsample block
- 参数
scale_factor (
int
) – Upsample scale factor.
- class basecls.models.hrnet.HRFusion(channels, multi_scale_output, norm_name, act_name)[源代码]#
基类:
Module
HRNet fusion block.
- 参数
- class basecls.models.hrnet.HRModule(block_name, num_blocks, in_channels, channels, multi_scale_output, norm_name, act_name)[源代码]#
基类:
Module
HRNet module.
- class basecls.models.hrnet.HRTrans(in_chs, out_chs, norm_name, act_name)[源代码]#
基类:
Module
HRNet transition block.
- 参数
- class basecls.models.hrnet.HRStage(num_modules, num_blocks, block_name, pre_channels, cur_channels, multi_scale_output, w_fst, norm_name, act_name)[源代码]#
基类:
Module
HRNet stage.
- 参数
num_modules (
int
) – Number of modules.block_name (
str
) – Branch block type.pre_channels (
List
[int
]) – Channels of previous stage (an empty list for the first stage).multi_scale_output (
bool
) – Whether output multi-scale features.w_fst (
Optional
[int
]) – Width of stem for the first stage (None
for other stages).norm_name (
str
) – Normalization layer.act_name (
str
) – Activation function.
- class basecls.models.hrnet.HRMerge(block_name, pre_channels, channels, norm_name, act_name)[源代码]#
基类:
Module
HRNet merge block.
- 参数
- class basecls.models.hrnet.HRNet(stage_modules, stage_blocks, stage_block_names, stage_channels, w_stem=64, multi_scale_output=True, merge_block_name='bottleneck', merge_channels=[32, 64, 128, 256], norm_name='BN', act_name='relu', head=None, **kwargs)[源代码]#
基类:
Module
HRNet model.
- 参数
stage_modules (
List
[int
]) – Number of modules for each stage.stage_blocks (
List
[List
[int
]]) – Number of blocks for each module in stages.stage_block_names (
List
[str
]) – Branch block types for each stage.stage_channels (
List
[List
[int
]]) – Number of channels for each stage.w_stem (
int
) – Stem width. Default:64
multi_scale_output (
bool
) – Whether output multi-scale features. Default:True
merge_block_name (
str
) – Merge block type. Default:"bottleneck"
merge_channels (
List
[int
]) – Channels of each scale in merge block. Default:[32, 64, 128, 256]
norm_name (
str
) – Normalization layer. Default:"BN"
act_name (
str
) – Activation function. Default:"relu"
head (
Optional
[Mapping
[str
,Any
]]) – head args. Default:None