esnet.py

# 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
from __future__ import division
from __future__ import print_function

import paddle
import paddle.nn as nn
import paddle.nn.functional as F
from paddle import ParamAttr
from paddle.nn import Conv2D, MaxPool2D, AdaptiveAvgPool2D, BatchNorm
from paddle.nn.initializer import KaimingNormal
from paddle.regularizer import L2Decay

from ppdet.core.workspace import register, serializable
from numbers import Integral
from ..shape_spec import ShapeSpec
from ppdet.modeling.ops import channel_shuffle
from ppdet.modeling.backbones.shufflenet_v2 import ConvBNLayer

__all__ = ['ESNet']


def make_divisible(v, divisor=16, 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 SEModule(nn.Layer):
    def __init__(self, channel, reduction=4):
        super(SEModule, self).__init__()
        self.avg_pool = AdaptiveAvgPool2D(1)
        self.conv1 = Conv2D(
            in_channels=channel,
            out_channels=channel // reduction,
            kernel_size=1,
            stride=1,
            padding=0,
            weight_attr=ParamAttr(),
            bias_attr=ParamAttr())
        self.conv2 = Conv2D(
            in_channels=channel // reduction,
            out_channels=channel,
            kernel_size=1,
            stride=1,
            padding=0,
            weight_attr=ParamAttr(),
            bias_attr=ParamAttr())

    def forward(self, inputs):
        outputs = self.avg_pool(inputs)
        outputs = self.conv1(outputs)
        outputs = F.relu(outputs)
        outputs = self.conv2(outputs)
        outputs = F.hardsigmoid(outputs)
        return paddle.multiply(x=inputs, y=outputs)


class InvertedResidual(nn.Layer):
    def __init__(self,
                 in_channels,
                 mid_channels,
                 out_channels,
                 stride,
                 act="relu"):
        super(InvertedResidual, self).__init__()
        self._conv_pw = ConvBNLayer(
            in_channels=in_channels // 2,
            out_channels=mid_channels // 2,
            kernel_size=1,
            stride=1,
            padding=0,
            groups=1,
            act=act)
        self._conv_dw = ConvBNLayer(
            in_channels=mid_channels // 2,
            out_channels=mid_channels // 2,
            kernel_size=3,
            stride=stride,
            padding=1,
            groups=mid_channels // 2,
            act=None)
        self._se = SEModule(mid_channels)

        self._conv_linear = ConvBNLayer(
            in_channels=mid_channels,
            out_channels=out_channels // 2,
            kernel_size=1,
            stride=1,
            padding=0,
            groups=1,
            act=act)

    def forward(self, inputs):
        x1, x2 = paddle.split(
            inputs,
            num_or_sections=[inputs.shape[1] // 2, inputs.shape[1] // 2],
            axis=1)
        x2 = self._conv_pw(x2)
        x3 = self._conv_dw(x2)
        x3 = paddle.concat([x2, x3], axis=1)
        x3 = self._se(x3)
        x3 = self._conv_linear(x3)
        out = paddle.concat([x1, x3], axis=1)
        return channel_shuffle(out, 2)


class InvertedResidualDS(nn.Layer):
    def __init__(self,
                 in_channels,
                 mid_channels,
                 out_channels,
                 stride,
                 act="relu"):
        super(InvertedResidualDS, self).__init__()

        # branch1
        self._conv_dw_1 = ConvBNLayer(
            in_channels=in_channels,
            out_channels=in_channels,
            kernel_size=3,
            stride=stride,
            padding=1,
            groups=in_channels,
            act=None)
        self._conv_linear_1 = ConvBNLayer(
            in_channels=in_channels,
            out_channels=out_channels // 2,
            kernel_size=1,
            stride=1,
            padding=0,
            groups=1,
            act=act)
        # branch2
        self._conv_pw_2 = ConvBNLayer(
            in_channels=in_channels,
            out_channels=mid_channels // 2,
            kernel_size=1,
            stride=1,
            padding=0,
            groups=1,
            act=act)
        self._conv_dw_2 = ConvBNLayer(
            in_channels=mid_channels // 2,
            out_channels=mid_channels // 2,
            kernel_size=3,
            stride=stride,
            padding=1,
            groups=mid_channels // 2,
            act=None)
        self._se = SEModule(mid_channels // 2)
        self._conv_linear_2 = ConvBNLayer(
            in_channels=mid_channels // 2,
            out_channels=out_channels // 2,
            kernel_size=1,
            stride=1,
            padding=0,
            groups=1,
            act=act)
        self._conv_dw_mv1 = ConvBNLayer(
            in_channels=out_channels,
            out_channels=out_channels,
            kernel_size=3,
            stride=1,
            padding=1,
            groups=out_channels,
            act="hard_swish")
        self._conv_pw_mv1 = ConvBNLayer(
            in_channels=out_channels,
            out_channels=out_channels,
            kernel_size=1,
            stride=1,
            padding=0,
            groups=1,
            act="hard_swish")

    def forward(self, inputs):
        x1 = self._conv_dw_1(inputs)
        x1 = self._conv_linear_1(x1)
        x2 = self._conv_pw_2(inputs)
        x2 = self._conv_dw_2(x2)
        x2 = self._se(x2)
        x2 = self._conv_linear_2(x2)
        out = paddle.concat([x1, x2], axis=1)
        out = self._conv_dw_mv1(out)
        out = self._conv_pw_mv1(out)

        return out


@register
@serializable
class ESNet(nn.Layer):
    def __init__(self,
                 scale=1.0,
                 act="hard_swish",
                 feature_maps=[4, 11, 14],
                 channel_ratio=[1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1]):
        super(ESNet, self).__init__()
        self.scale = scale
        if isinstance(feature_maps, Integral):
            feature_maps = [feature_maps]
        self.feature_maps = feature_maps
        stage_repeats = [3, 7, 3]

        stage_out_channels = [
            -1, 24, make_divisible(128 * scale), make_divisible(256 * scale),
            make_divisible(512 * scale), 1024
        ]

        self._out_channels = []
        self._feature_idx = 0
        # 1. conv1
        self._conv1 = ConvBNLayer(
            in_channels=3,
            out_channels=stage_out_channels[1],
            kernel_size=3,
            stride=2,
            padding=1,
            act=act)
        self._max_pool = MaxPool2D(kernel_size=3, stride=2, padding=1)
        self._feature_idx += 1

        # 2. bottleneck sequences
        self._block_list = []
        arch_idx = 0
        for stage_id, num_repeat in enumerate(stage_repeats):
            for i in range(num_repeat):
                channels_scales = channel_ratio[arch_idx]
                mid_c = make_divisible(
                    int(stage_out_channels[stage_id + 2] * channels_scales),
                    divisor=8)
                if i == 0:
                    block = self.add_sublayer(
                        name=str(stage_id + 2) + '_' + str(i + 1),
                        sublayer=InvertedResidualDS(
                            in_channels=stage_out_channels[stage_id + 1],
                            mid_channels=mid_c,
                            out_channels=stage_out_channels[stage_id + 2],
                            stride=2,
                            act=act))
                else:
                    block = self.add_sublayer(
                        name=str(stage_id + 2) + '_' + str(i + 1),
                        sublayer=InvertedResidual(
                            in_channels=stage_out_channels[stage_id + 2],
                            mid_channels=mid_c,
                            out_channels=stage_out_channels[stage_id + 2],
                            stride=1,
                            act=act))
                self._block_list.append(block)
                arch_idx += 1
                self._feature_idx += 1
                self._update_out_channels(stage_out_channels[stage_id + 2],
                                          self._feature_idx, self.feature_maps)

    def _update_out_channels(self, channel, feature_idx, feature_maps):
        if feature_idx in feature_maps:
            self._out_channels.append(channel)

    def forward(self, inputs):
        y = self._conv1(inputs['image'])
        y = self._max_pool(y)
        outs = []
        for i, inv in enumerate(self._block_list):
            y = inv(y)
            if i + 2 in self.feature_maps:
                outs.append(y)

        return outs

    @property
    def out_shape(self):
        return [ShapeSpec(channels=c) for c in self._out_channels]