add Rethinking BiSeNet (STDCSeg) to paddleseg

configs/stdcseg/README.md

@@ -0,0 +1,14 @@
+# Rethinking BiSeNet For Real-time Semantic Segmentation
+
+## Reference
+
+> Fan, Mingyuan, et al. "Rethinking BiSeNet For Real-time Semantic Segmentation." Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2021.
+
+
+## Performance
+
+### CityScapes
+
+| Model | Backbone | Resolution | Training Iters | mIoU | mIoU (flip) | mIoU (ms+flip) | Links |
+|---|---|---|---|---|---|---|---|
+|STDC2-Seg50|STDC1446|1024x512|80000|74.62%|-|-|[backbone提取码：tss7](https://pan.baidu.com/s/16kh3aHTBBX6wfKiIG-y3yA) [model+log提取码：nchx](https://pan.baidu.com/s/1sFHqZWhcl8hFzGCrXu_c7Q) [vdl](https://www.paddlepaddle.org.cn/paddle/visualdl/service/app/index?id=30a6031fcc7cc09db93b4d33eb21724a) |

configs/stdcseg/stdc2_seg_cityscapes_1024x512_80k.yml

@@ -0,0 +1,60 @@
+_base_: '../_base_/cityscapes.yml'
+
+batch_size: 36
+iters: 80000
+
+model:
+  type: STDCSeg
+  backbone:
+    type: STDC2
+    pretrained: '/home/path/STDCNet1446_76.47.pdiparams'
+  num_classes: 19
+  pretrained: null
+
+train_dataset:
+  type: Cityscapes
+  dataset_root: data/cityscapes
+  transforms:
+    - type: ResizeStepScaling
+      min_scale_factor: 0.125
+      max_scale_factor: 1.5
+      scale_step_size: 0.125
+    - type: RandomPaddingCrop
+      crop_size: [1024, 512]
+    - type: RandomHorizontalFlip
+    - type: RandomDistort
+      brightness_range: 0.5
+      contrast_range: 0.5
+      saturation_range: 0.5
+    - type: Normalize
+      mean: [0.485, 0.456, 0.406]
+      std: [0.229, 0.224, 0.225]
+  mode: train
+
+val_dataset:
+  type: Cityscapes
+  dataset_root: data/cityscapes
+  transforms:
+    - type: Normalize
+      mean: [0.485, 0.456, 0.406]
+      std: [0.229, 0.224, 0.225]
+  mode: val
+
+optimizer:
+  type: sgd
+  momentum: 0.9
+  weight_decay: 4.0e-5
+
+loss:
+  types:
+    - type: OhemCrossEntropyLoss
+    - type: OhemCrossEntropyLoss
+    - type: OhemCrossEntropyLoss
+    - type: DetailAggregateLoss
+  coef: [1, 1, 1, 1]
+
+lr_scheduler:
+  type: PolynomialDecay
+  learning_rate: 0.01
+  end_lr: 0
+  power: 0.9

paddleseg/models/__init__.py

@@ -40,4 +40,5 @@
 from .ppseg_lite import *
 from .mla_transformer import MLATransformer
 from .portraitnet import PortraitNet
+from .stdcseg import STDCSeg
 from .segformer import SegFormer

paddleseg/models/backbones/__init__.py

@@ -20,3 +20,4 @@
 from .swin_transformer import *
 from .mobilenetv2 import *
 from .mix_transformer import *
+from .stdcnet import *

paddleseg/models/backbones/stdcnet.py

@@ -0,0 +1,247 @@
+# Copyright (c) 2021 PaddlePaddle Authors. All Rights Reserved.
+#
+# 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.
+
+import math
+
+import paddle
+import paddle.nn as nn
+
+from paddleseg.utils import utils
+from paddleseg.cvlibs import manager,param_init
+from paddleseg.models.layers import ConvBNReLU
+
+__all__ = ["STDC1", "STDC2"]
+
+
+class STDCNet(nn.Layer):
+    """
+    The STDCNet implementation based on PaddlePaddle.
+
+    The original article refers to Meituan
+    Fan, Mingyuan, et al. "Rethinking BiSeNet For Real-time Semantic Segmentation."
+    (https://arxiv.org/abs/2104.13188)
+
+    Args:
+        base(int, optional): base channels. Default: 64.
+        layers(list, optional): layers numbers list. It determines STDC block numbers of STDCNet's stage3\4\5. Defualt: [4, 5, 3].
+        block_num(int,optional): block_num of features block. Default: 4.
+        type(str,optional): feature fusion method "cat"/"add". Default: "cat".
+        num_classes(int, optional): class number for image classification. Default: 1000.
+        dropout(float,optional): dropout ratio. if >0,use dropout ratio.  Default: 0.20.
+        use_conv_last(bool,optional): whether to use the last ConvBNReLU layer . Default: False.
+        pretrained(str, optional): the path of pretrained model.
+    """
+
+    def __init__(self, base=64,
+                 layers=[4, 5, 3],
+                 block_num=4,
+                 type="cat",
+                 num_classes=1000,
+                 dropout=0.20,
+                 use_conv_last=False,
+                 pretrained=None):
+        super(STDCNet, self).__init__()
+        if type == "cat":
+            block = CatBottleneck
+        elif type == "add":
+            block = AddBottleneck
+        self.use_conv_last = use_conv_last
+        self.features = self._make_layers(base, layers, block_num, block)
+        self.conv_last = ConvBNReLU(base * 16, max(1024, base * 16), kernel_size=1,stride=1,padding=0)
+        self.gap = nn.AdaptiveAvgPool2D(1)
+        self.fc = nn.Linear(max(1024, base * 16), max(1024, base * 16),bias_attr=None)
+        self.bn = nn.BatchNorm1D(max(1024, base * 16))
+        self.relu = nn.ReLU()
+        self.dropout = nn.Dropout(p=dropout)
+        self.linear = nn.Linear(max(1024, base * 16), num_classes, bias_attr=None)
+
+        if(layers==[4,5,3]): #stdc1446
+            self.x2 = nn.Sequential(self.features[:1])
+            self.x4 = nn.Sequential(self.features[1:2])
+            self.x8 = nn.Sequential(self.features[2:6])
+            self.x16 = nn.Sequential(self.features[6:11])
+            self.x32 = nn.Sequential(self.features[11:])
+        elif(layers==[2,2,2]):#stdc813
+            self.x2 = nn.Sequential(self.features[:1])
+            self.x4 = nn.Sequential(self.features[1:2])
+            self.x8 = nn.Sequential(self.features[2:4])
+            self.x16 = nn.Sequential(self.features[4:6])
+            self.x32 = nn.Sequential(self.features[6:])
+        else:
+            raise NotImplementedError("model with layers:{} is not implemented!".format(layers))
+
+        self.pretrained = pretrained
+        self.init_weight()
+
+    def forward(self, x):
+        """
+        forward function for feature extract.
+        """
+        feat2 = self.x2(x)
+        feat4 = self.x4(feat2)
+        feat8 = self.x8(feat4)
+        feat16 = self.x16(feat8)
+        feat32 = self.x32(feat16)
+        if self.use_conv_last:
+            feat32 = self.conv_last(feat32)
+        return feat2, feat4, feat8, feat16, feat32
+
+    def forward_impl(self, x):
+        """
+        forward function for classification.
+        """
+        out = self.features(x)
+        out = self.conv_last(out).pow(2)
+        out = self.gap(out).flatten(1)
+        out = self.fc(out)
+        out = self.relu(out)
+        out = self.dropout(out)
+        out = self.linear(out)
+        return out
+
+    def _make_layers(self, base, layers, block_num, block):
+        features = []
+        features += [ConvBNReLU(3, base // 2, kernel_size=3, stride=2,padding=1)]
+        features += [ConvBNReLU(base // 2, base, kernel_size=3, stride=2,padding=1)]
+
+        for i, layer in enumerate(layers):
+            for j in range(layer):
+                if i == 0 and j == 0:
+                    features.append(block(base, base * 4, block_num, 2))
+                elif j == 0:
+                    features.append(block(base * int(math.pow(2, i + 1)), base * int(math.pow(2, i + 2)), block_num, 2))
+                else:
+                    features.append(block(base * int(math.pow(2, i + 2)), base * int(math.pow(2, i + 2)), block_num, 1))
+
+        return nn.Sequential(*features)
+
+    def init_weight(self):
+        for layer in self.sublayers():
+            if isinstance(layer, nn.Conv2D):
+                param_init.normal_init(layer.weight, std=0.001)
+            elif isinstance(layer, (nn.BatchNorm, nn.SyncBatchNorm)):
+                param_init.constant_init(layer.weight, value=1.0)
+                param_init.constant_init(layer.bias, value=0.0)
+        if self.pretrained is not None:
+            utils.load_pretrained_model(self, self.pretrained)
+
+
+class AddBottleneck(nn.Layer):
+    def __init__(self, in_planes, out_planes, block_num=3, stride=1):
+        super(AddBottleneck, self).__init__()
+        assert block_num > 1, print("block number should be larger than 1.")
+        self.conv_list = nn.LayerList()
+        self.stride = stride
+
+        if stride == 2:
+            self.avd_layer = nn.Sequential(
+                nn.Conv2D(out_planes // 2, out_planes // 2, kernel_size=3, stride=2, padding=1, groups=out_planes // 2,bias_attr=None),
+                nn.BatchNorm2D(out_planes // 2),
+            )
+            self.skip = nn.Sequential(
+                nn.Conv2D(in_planes, in_planes, kernel_size=3, stride=2, padding=1, groups=in_planes,bias_attr=None),
+                nn.BatchNorm2D(in_planes),
+                nn.Conv2D(in_planes, out_planes, kernel_size=1,bias_attr=None),
+                nn.BatchNorm2D(out_planes),
+            )
+            stride = 1
+
+        for idx in range(block_num):
+            if idx == 0:
+                self.conv_list.append(ConvBNReLU(in_planes, out_planes // 2, kernel_size=1, stride=1, padding=0, bias_attr=None))
+            elif idx == 1 and block_num == 2:
+                self.conv_list.append(ConvBNReLU(out_planes // 2, out_planes // 2, kernel_size=3, stride=stride, padding=1, bias_attr=None))
+            elif idx == 1 and block_num > 2:
+                self.conv_list.append(ConvBNReLU(out_planes // 2, out_planes // 4, kernel_size=3, stride=stride,padding=1, bias_attr=None))
+            elif idx < block_num - 1:
+                self.conv_list.append(
+                    ConvBNReLU(out_planes // int(math.pow(2, idx)), out_planes // int(math.pow(2, idx + 1)), kernel_size=3,
+                               stride=1, padding=1, bias_attr=None))
+            else:
+                self.conv_list.append(ConvBNReLU(out_planes // int(math.pow(2, idx)), out_planes // int(math.pow(2, idx)),
+                                                 kernel_size=3, stride=1, padding=1, bias_attr=None))
+
+    def forward(self, x):
+        out_list = []
+        out = x
+        for idx, conv in enumerate(self.conv_list):
+            if idx == 0 and self.stride == 2:
+                out = self.avd_layer(conv(out))
+            else:
+                out = conv(out)
+            out_list.append(out)
+        if self.stride == 2:
+            x = self.skip(x)
+        return paddle.concat(out_list, axis=1) + x
+
+
+class CatBottleneck(nn.Layer):
+    def __init__(self, in_planes, out_planes, block_num=3, stride=1):
+        super(CatBottleneck, self).__init__()
+        assert block_num > 1, print("block number should be larger than 1.")
+        self.conv_list = nn.LayerList()
+        self.stride = stride
+
+        if stride == 2:
+            self.avd_layer = nn.Sequential(
+                nn.Conv2D(out_planes // 2, out_planes // 2, kernel_size=3, stride=2, padding=1, groups=out_planes // 2,bias_attr=None),
+                nn.BatchNorm2D(out_planes // 2),
+            )
+            self.skip = nn.AvgPool2D(kernel_size=3, stride=2, padding=1)
+            stride = 1
+
+        for idx in range(block_num):
+            if idx == 0:
+                self.conv_list.append(ConvBNReLU(in_planes, out_planes // 2, kernel_size=1, stride=1, padding=0, bias_attr=None))
+            elif idx == 1 and block_num == 2:
+                self.conv_list.append(ConvBNReLU(out_planes // 2, out_planes // 2, kernel_size=3, stride=stride, padding=1, bias_attr=None))
+            elif idx == 1 and block_num > 2:
+                self.conv_list.append(ConvBNReLU(out_planes // 2, out_planes // 4, kernel_size=3, stride=stride, padding=1, bias_attr=None))
+            elif idx < block_num - 1:
+                self.conv_list.append(
+                    ConvBNReLU(out_planes // int(math.pow(2, idx)), out_planes // int(math.pow(2, idx + 1)), kernel_size=3,
+                               stride=1, padding=1, bias_attr=None))
+            else:
+                self.conv_list.append(ConvBNReLU(out_planes // int(math.pow(2, idx)), out_planes // int(math.pow(2, idx)),
+                                                 kernel_size=3, stride=1, padding=1, bias_attr=None))
+
+    def forward(self, x):
+        out_list = []
+        out1 = self.conv_list[0](x)
+        for idx, conv in enumerate(self.conv_list[1:]):
+            if idx == 0:
+                if self.stride == 2:
+                    out = conv(self.avd_layer(out1))
+                else:
+                    out = conv(out1)
+            else:
+                out = conv(out)
+            out_list.append(out)
+
+        if self.stride == 2:
+            out1 = self.skip(out1)
+        out_list.insert(0, out1)
+        out = paddle.concat(out_list, axis=1)
+        return out
+
+
+@manager.BACKBONES.add_component
+def STDC2(**kwargs):
+    model = STDCNet(base=64,layers=[4,5,3],**kwargs)
+    return model
+
+@manager.BACKBONES.add_component
+def STDC1(**kwargs):
+    model = STDCNet(base=64,layers=[2,2,2],**kwargs)
+    return model

paddleseg/models/losses/__init__.py

@@ -29,3 +29,4 @@
 from .focal_loss import FocalLoss
 from .kl_loss import KLLoss
 from .rmi_loss import RMILoss
+from .detail_aggregate_loss import DetailAggregateLoss