File size: 5,992 Bytes
a89d9fd |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 |
# copyright (c) 2019 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
from paddle import nn
import paddle.nn.functional as F
from paddle import ParamAttr
import os
import sys
import math
from paddle.nn.initializer import TruncatedNormal, Constant, Normal
ones_ = Constant(value=1.)
zeros_ = Constant(value=0.)
__dir__ = os.path.dirname(os.path.abspath(__file__))
sys.path.append(__dir__)
sys.path.insert(0, os.path.abspath(os.path.join(__dir__, '../../..')))
class Conv_BN_ReLU(nn.Layer):
def __init__(self,
in_planes,
out_planes,
kernel_size=1,
stride=1,
padding=0):
super(Conv_BN_ReLU, self).__init__()
self.conv = nn.Conv2D(
in_planes,
out_planes,
kernel_size=kernel_size,
stride=stride,
padding=padding,
bias_attr=False)
self.bn = nn.BatchNorm2D(out_planes)
self.relu = nn.ReLU()
for m in self.sublayers():
if isinstance(m, nn.Conv2D):
n = m._kernel_size[0] * m._kernel_size[1] * m._out_channels
normal_ = Normal(mean=0.0, std=math.sqrt(2. / n))
normal_(m.weight)
elif isinstance(m, nn.BatchNorm2D):
zeros_(m.bias)
ones_(m.weight)
def forward(self, x):
return self.relu(self.bn(self.conv(x)))
class FPEM(nn.Layer):
def __init__(self, in_channels, out_channels):
super(FPEM, self).__init__()
planes = out_channels
self.dwconv3_1 = nn.Conv2D(
planes,
planes,
kernel_size=3,
stride=1,
padding=1,
groups=planes,
bias_attr=False)
self.smooth_layer3_1 = Conv_BN_ReLU(planes, planes)
self.dwconv2_1 = nn.Conv2D(
planes,
planes,
kernel_size=3,
stride=1,
padding=1,
groups=planes,
bias_attr=False)
self.smooth_layer2_1 = Conv_BN_ReLU(planes, planes)
self.dwconv1_1 = nn.Conv2D(
planes,
planes,
kernel_size=3,
stride=1,
padding=1,
groups=planes,
bias_attr=False)
self.smooth_layer1_1 = Conv_BN_ReLU(planes, planes)
self.dwconv2_2 = nn.Conv2D(
planes,
planes,
kernel_size=3,
stride=2,
padding=1,
groups=planes,
bias_attr=False)
self.smooth_layer2_2 = Conv_BN_ReLU(planes, planes)
self.dwconv3_2 = nn.Conv2D(
planes,
planes,
kernel_size=3,
stride=2,
padding=1,
groups=planes,
bias_attr=False)
self.smooth_layer3_2 = Conv_BN_ReLU(planes, planes)
self.dwconv4_2 = nn.Conv2D(
planes,
planes,
kernel_size=3,
stride=2,
padding=1,
groups=planes,
bias_attr=False)
self.smooth_layer4_2 = Conv_BN_ReLU(planes, planes)
def _upsample_add(self, x, y):
return F.upsample(x, scale_factor=2, mode='bilinear') + y
def forward(self, f1, f2, f3, f4):
# up-down
f3 = self.smooth_layer3_1(self.dwconv3_1(self._upsample_add(f4, f3)))
f2 = self.smooth_layer2_1(self.dwconv2_1(self._upsample_add(f3, f2)))
f1 = self.smooth_layer1_1(self.dwconv1_1(self._upsample_add(f2, f1)))
# down-up
f2 = self.smooth_layer2_2(self.dwconv2_2(self._upsample_add(f2, f1)))
f3 = self.smooth_layer3_2(self.dwconv3_2(self._upsample_add(f3, f2)))
f4 = self.smooth_layer4_2(self.dwconv4_2(self._upsample_add(f4, f3)))
return f1, f2, f3, f4
class CTFPN(nn.Layer):
def __init__(self, in_channels, out_channel=128):
super(CTFPN, self).__init__()
self.out_channels = out_channel * 4
self.reduce_layer1 = Conv_BN_ReLU(in_channels[0], 128)
self.reduce_layer2 = Conv_BN_ReLU(in_channels[1], 128)
self.reduce_layer3 = Conv_BN_ReLU(in_channels[2], 128)
self.reduce_layer4 = Conv_BN_ReLU(in_channels[3], 128)
self.fpem1 = FPEM(in_channels=(64, 128, 256, 512), out_channels=128)
self.fpem2 = FPEM(in_channels=(64, 128, 256, 512), out_channels=128)
def _upsample(self, x, scale=1):
return F.upsample(x, scale_factor=scale, mode='bilinear')
def forward(self, f):
# # reduce channel
f1 = self.reduce_layer1(f[0]) # N,64,160,160 --> N, 128, 160, 160
f2 = self.reduce_layer2(f[1]) # N, 128, 80, 80 --> N, 128, 80, 80
f3 = self.reduce_layer3(f[2]) # N, 256, 40, 40 --> N, 128, 40, 40
f4 = self.reduce_layer4(f[3]) # N, 512, 20, 20 --> N, 128, 20, 20
# FPEM
f1_1, f2_1, f3_1, f4_1 = self.fpem1(f1, f2, f3, f4)
f1_2, f2_2, f3_2, f4_2 = self.fpem2(f1_1, f2_1, f3_1, f4_1)
# FFM
f1 = f1_1 + f1_2
f2 = f2_1 + f2_2
f3 = f3_1 + f3_2
f4 = f4_1 + f4_2
f2 = self._upsample(f2, scale=2)
f3 = self._upsample(f3, scale=4)
f4 = self._upsample(f4, scale=8)
ff = paddle.concat((f1, f2, f3, f4), 1) # N,512, 160,160
return ff
|