| 123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355 |
- # copyright (c) 2021 PaddlePaddle Authors. All Rights Reserve.
- #
- # Licensed under the Apache License, Version 2.0 (the "License");
- # you may not use this file except in compliance with the License.
- # You may obtain a copy of the License at
- #
- # http://www.apache.org/licenses/LICENSE-2.0
- #
- # Unless required by applicable law or agreed to in writing, software
- # distributed under the License is distributed on an "AS IS" BASIS,
- # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
- # See the License for the specific language governing permissions and
- # limitations under the License.
- from __future__ import absolute_import, division, print_function
- import math
- import paddle
- from paddle import ParamAttr, reshape, transpose, concat, split
- import paddle.nn as nn
- from paddle.nn import Conv2D, BatchNorm, Linear, Dropout
- from paddle.nn import AdaptiveAvgPool2D, MaxPool2D
- from paddle.nn.initializer import KaimingNormal
- from paddle.regularizer import L2Decay
- from paddlex.ppcls.arch.backbone.base.theseus_layer import TheseusLayer
- from paddlex.ppcls.utils.save_load import load_dygraph_pretrain, load_dygraph_pretrain_from_url
- MODEL_URLS = {
- "ESNet_x0_25":
- "https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/ESNet_x0_25_pretrained.pdparams",
- "ESNet_x0_5":
- "https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/ESNet_x0_5_pretrained.pdparams",
- "ESNet_x0_75":
- "https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/ESNet_x0_75_pretrained.pdparams",
- "ESNet_x1_0":
- "https://paddle-imagenet-models-name.bj.bcebos.com/dygraph/legendary_models/ESNet_x1_0_pretrained.pdparams",
- }
- __all__ = list(MODEL_URLS.keys())
- def channel_shuffle(x, groups):
- batch_size, num_channels, height, width = x.shape[0:4]
- channels_per_group = num_channels // groups
- x = reshape(
- x=x, shape=[batch_size, groups, channels_per_group, height, width])
- x = transpose(x=x, perm=[0, 2, 1, 3, 4])
- x = reshape(x=x, shape=[batch_size, num_channels, height, width])
- return x
- def make_divisible(v, divisor=8, min_value=None):
- if min_value is None:
- min_value = divisor
- new_v = max(min_value, int(v + divisor / 2) // divisor * divisor)
- if new_v < 0.9 * v:
- new_v += divisor
- return new_v
- class ConvBNLayer(TheseusLayer):
- def __init__(self,
- in_channels,
- out_channels,
- kernel_size,
- stride=1,
- groups=1,
- if_act=True):
- super().__init__()
- self.conv = Conv2D(
- in_channels=in_channels,
- out_channels=out_channels,
- kernel_size=kernel_size,
- stride=stride,
- padding=(kernel_size - 1) // 2,
- groups=groups,
- weight_attr=ParamAttr(initializer=KaimingNormal()),
- bias_attr=False)
- self.bn = BatchNorm(
- out_channels,
- param_attr=ParamAttr(regularizer=L2Decay(0.0)),
- bias_attr=ParamAttr(regularizer=L2Decay(0.0)))
- self.if_act = if_act
- self.hardswish = nn.Hardswish()
- def forward(self, x):
- x = self.conv(x)
- x = self.bn(x)
- if self.if_act:
- x = self.hardswish(x)
- return x
- class SEModule(TheseusLayer):
- def __init__(self, channel, reduction=4):
- super().__init__()
- self.avg_pool = AdaptiveAvgPool2D(1)
- self.conv1 = Conv2D(
- in_channels=channel,
- out_channels=channel // reduction,
- kernel_size=1,
- stride=1,
- padding=0)
- self.relu = nn.ReLU()
- self.conv2 = Conv2D(
- in_channels=channel // reduction,
- out_channels=channel,
- kernel_size=1,
- stride=1,
- padding=0)
- self.hardsigmoid = nn.Hardsigmoid()
- def forward(self, x):
- identity = x
- x = self.avg_pool(x)
- x = self.conv1(x)
- x = self.relu(x)
- x = self.conv2(x)
- x = self.hardsigmoid(x)
- x = paddle.multiply(x=identity, y=x)
- return x
- class ESBlock1(TheseusLayer):
- def __init__(self, in_channels, out_channels):
- super().__init__()
- self.pw_1_1 = ConvBNLayer(
- in_channels=in_channels // 2,
- out_channels=out_channels // 2,
- kernel_size=1,
- stride=1)
- self.dw_1 = ConvBNLayer(
- in_channels=out_channels // 2,
- out_channels=out_channels // 2,
- kernel_size=3,
- stride=1,
- groups=out_channels // 2,
- if_act=False)
- self.se = SEModule(out_channels)
- self.pw_1_2 = ConvBNLayer(
- in_channels=out_channels,
- out_channels=out_channels // 2,
- kernel_size=1,
- stride=1)
- def forward(self, x):
- x1, x2 = split(
- x, num_or_sections=[x.shape[1] // 2, x.shape[1] // 2], axis=1)
- x2 = self.pw_1_1(x2)
- x3 = self.dw_1(x2)
- x3 = concat([x2, x3], axis=1)
- x3 = self.se(x3)
- x3 = self.pw_1_2(x3)
- x = concat([x1, x3], axis=1)
- return channel_shuffle(x, 2)
- class ESBlock2(TheseusLayer):
- def __init__(self, in_channels, out_channels):
- super().__init__()
- # branch1
- self.dw_1 = ConvBNLayer(
- in_channels=in_channels,
- out_channels=in_channels,
- kernel_size=3,
- stride=2,
- groups=in_channels,
- if_act=False)
- self.pw_1 = ConvBNLayer(
- in_channels=in_channels,
- out_channels=out_channels // 2,
- kernel_size=1,
- stride=1)
- # branch2
- self.pw_2_1 = ConvBNLayer(
- in_channels=in_channels,
- out_channels=out_channels // 2,
- kernel_size=1)
- self.dw_2 = ConvBNLayer(
- in_channels=out_channels // 2,
- out_channels=out_channels // 2,
- kernel_size=3,
- stride=2,
- groups=out_channels // 2,
- if_act=False)
- self.se = SEModule(out_channels // 2)
- self.pw_2_2 = ConvBNLayer(
- in_channels=out_channels // 2,
- out_channels=out_channels // 2,
- kernel_size=1)
- self.concat_dw = ConvBNLayer(
- in_channels=out_channels,
- out_channels=out_channels,
- kernel_size=3,
- groups=out_channels)
- self.concat_pw = ConvBNLayer(
- in_channels=out_channels, out_channels=out_channels, kernel_size=1)
- def forward(self, x):
- x1 = self.dw_1(x)
- x1 = self.pw_1(x1)
- x2 = self.pw_2_1(x)
- x2 = self.dw_2(x2)
- x2 = self.se(x2)
- x2 = self.pw_2_2(x2)
- x = concat([x1, x2], axis=1)
- x = self.concat_dw(x)
- x = self.concat_pw(x)
- return x
- class ESNet(TheseusLayer):
- def __init__(self,
- class_num=1000,
- scale=1.0,
- dropout_prob=0.2,
- class_expand=1280):
- super().__init__()
- self.scale = scale
- self.class_num = class_num
- self.class_expand = class_expand
- stage_repeats = [3, 7, 3]
- stage_out_channels = [
- -1, 24, make_divisible(116 * scale), make_divisible(232 * scale),
- make_divisible(464 * scale), 1024
- ]
- self.conv1 = ConvBNLayer(
- in_channels=3,
- out_channels=stage_out_channels[1],
- kernel_size=3,
- stride=2)
- self.max_pool = MaxPool2D(kernel_size=3, stride=2, padding=1)
- block_list = []
- for stage_id, num_repeat in enumerate(stage_repeats):
- for i in range(num_repeat):
- if i == 0:
- block = ESBlock2(
- in_channels=stage_out_channels[stage_id + 1],
- out_channels=stage_out_channels[stage_id + 2])
- else:
- block = ESBlock1(
- in_channels=stage_out_channels[stage_id + 2],
- out_channels=stage_out_channels[stage_id + 2])
- block_list.append(block)
- self.blocks = nn.Sequential(*block_list)
- self.conv2 = ConvBNLayer(
- in_channels=stage_out_channels[-2],
- out_channels=stage_out_channels[-1],
- kernel_size=1)
- self.avg_pool = AdaptiveAvgPool2D(1)
- self.last_conv = Conv2D(
- in_channels=stage_out_channels[-1],
- out_channels=self.class_expand,
- kernel_size=1,
- stride=1,
- padding=0,
- bias_attr=False)
- self.hardswish = nn.Hardswish()
- self.dropout = Dropout(p=dropout_prob, mode="downscale_in_infer")
- self.flatten = nn.Flatten(start_axis=1, stop_axis=-1)
- self.fc = Linear(self.class_expand, self.class_num)
- def forward(self, x):
- x = self.conv1(x)
- x = self.max_pool(x)
- x = self.blocks(x)
- x = self.conv2(x)
- x = self.avg_pool(x)
- x = self.last_conv(x)
- x = self.hardswish(x)
- x = self.dropout(x)
- x = self.flatten(x)
- x = self.fc(x)
- return x
- def _load_pretrained(pretrained, model, model_url, use_ssld):
- if pretrained is False:
- pass
- elif pretrained is True:
- load_dygraph_pretrain_from_url(model, model_url, use_ssld=use_ssld)
- elif isinstance(pretrained, str):
- load_dygraph_pretrain(model, pretrained)
- else:
- raise RuntimeError(
- "pretrained type is not available. Please use `string` or `boolean` type."
- )
- def ESNet_x0_25(pretrained=False, use_ssld=False, **kwargs):
- """
- ESNet_x0_25
- Args:
- pretrained: bool=False or str. If `True` load pretrained parameters, `False` otherwise.
- If str, means the path of the pretrained model.
- use_ssld: bool=False. Whether using distillation pretrained model when pretrained=True.
- Returns:
- model: nn.Layer. Specific `ESNet_x0_25` model depends on args.
- """
- model = ESNet(scale=0.25, **kwargs)
- _load_pretrained(pretrained, model, MODEL_URLS["ESNet_x0_25"], use_ssld)
- return model
- def ESNet_x0_5(pretrained=False, use_ssld=False, **kwargs):
- """
- ESNet_x0_5
- Args:
- pretrained: bool=False or str. If `True` load pretrained parameters, `False` otherwise.
- If str, means the path of the pretrained model.
- use_ssld: bool=False. Whether using distillation pretrained model when pretrained=True.
- Returns:
- model: nn.Layer. Specific `ESNet_x0_5` model depends on args.
- """
- model = ESNet(scale=0.5, **kwargs)
- _load_pretrained(pretrained, model, MODEL_URLS["ESNet_x0_5"], use_ssld)
- return model
- def ESNet_x0_75(pretrained=False, use_ssld=False, **kwargs):
- """
- ESNet_x0_75
- Args:
- pretrained: bool=False or str. If `True` load pretrained parameters, `False` otherwise.
- If str, means the path of the pretrained model.
- use_ssld: bool=False. Whether using distillation pretrained model when pretrained=True.
- Returns:
- model: nn.Layer. Specific `ESNet_x0_75` model depends on args.
- """
- model = ESNet(scale=0.75, **kwargs)
- _load_pretrained(pretrained, model, MODEL_URLS["ESNet_x0_75"], use_ssld)
- return model
- def ESNet_x1_0(pretrained=False, use_ssld=False, **kwargs):
- """
- ESNet_x1_0
- Args:
- pretrained: bool=False or str. If `True` load pretrained parameters, `False` otherwise.
- If str, means the path of the pretrained model.
- use_ssld: bool=False. Whether using distillation pretrained model when pretrained=True.
- Returns:
- model: nn.Layer. Specific `ESNet_x1_0` model depends on args.
- """
- model = ESNet(scale=1.0, **kwargs)
- _load_pretrained(pretrained, model, MODEL_URLS["ESNet_x1_0"], use_ssld)
- return model
|