语义分割系列22-GCNet（pytorch实现）

GCNet：《GCNet: Non-local Networks Meet Squeeze-Excitation Networks and Beyond》
论文链接：ICCV2019：GCNet

本文将介绍：

• GCNet: Non-local Networks Meet Squeeze-Excitation Networks and Beyond论文详解
• Non Local、Global context (GC) block的实现原理与区别
• Global context (GC) block的pytorch代码实现
• GCNet在Camvid数据集上的复现

---

---

引文

• 在以往的大量工作中证明，捕获视觉场景中的全局依赖能提高分割任务的效果。在传统的CNN网络工作中，远程依赖的建立（等同于感受野的扩增）主要依靠堆叠卷积层来实现，但是这种方法效率低且难以优化，因为长距离位置之间的信息难以传递，而且卷积层的堆叠可能会导致卷积核退化的问题。为了解决这个问题，Non-Local通过自注意力（self-attention）机制来建立远程依赖。对于每一个查询（query），计算该query位置与全局所有位置（key）的关系来建立注意力图（attention map），然后将注意力图与value进行加权汇总，生成最终的输出。

• Non-Local对比以往的工作而言（如叠加卷积、ASPP、PPM等等），其建立远程关系的能力十分优秀，但是Non-Local十分巨大的计算量成为了其进一步应用的缺陷。因此，后续一些工作也针对减少Non-Local的计算量开展了一些研究，包括Criss-Cross attention（CCNet）、（Asymmetric Non-local）ANNNet。而本文同样基于Non-local的计算量进行了优化。

本文主要工作：

• 对比了Non Local和简化版本的Non Local的计算效果
• 引入Context model到简化版本的Non Local结构中，提出Global Context（GC）block
  

论文主体

Non Local 与简化 Non Local 对比

图1 Non-Local与简化Non-Local结构对比 对于Non-Local (a)：

首先通过卷积计算出Key和Query :

$key=W_k(X)$ , $query=W_q(X)$
然后计算Matmul（K, Q）与SoftMax得到Attention map：
$Atten=SoftMax(key\odot query)$ ,
$query=W_q(X)$ 最后Attention map与Value进行计算Matmul（Attention，V），得到输出Out：
$Out=W_k（Atten\odot value）$

对于Simplified Non-Local结构，简化了query运算，query和key计算时权重共享，也就是query和key等同，这里就减少了计算query的一个过程，并忽略了最后一个卷积$W_z$，作者也给出了实际的证明——可视化了两种结构的attention map，发现效果差不多。因此作者后续的工作都基于Simplified Non-Local结构。

Global Context Block 细节与代码

作者基于Simplified Non Local 和SE block提出了Global Context（GC）block。

对于SE block的结构，主要由Context modeling构成，这个模块由SENet中所提出，后来在ENCNet中也有应用。在Context Modeling模块中，特征图通过一个全局平均池化，经由一系列卷积、ReLU之后，生成一个全局上下文特征（Context）（在ENCNet中，则是用这个全局上下文特征来实现通道注意力机制）。同时，这里也减少了许多计算量。

作者从SE block中得到灵感，将Context Modeling用在计算Attention map中，也就是用Context Modeling的输出来替代Attention Map，再接上卷积来实现value与Attention的输出，在这里，为了进一步简化计算作者设置了一个ratio(r=16)，缩减了通道数。

当然，这里show the code更为直接：

import torch
import torch.nn as nn
import torch.nn.functional as F
class GlobalContextBlock(nn.Module):
    def __init__(self, in_channels, scale = 16):
        super(GlobalContextBlock, self).__init__()
        self.in_channels = in_channels
        self.out_channels = self.in_channels//scale

        self.Conv_key = nn.Conv2d(self.in_channels, 1, 1)
        self.SoftMax = nn.Softmax(dim=1)

        self.Conv_value = nn.Sequential(
            nn.Conv2d(self.in_channels, self.out_channels, 1),
            nn.LayerNorm([self.out_channels, 1, 1]),
            nn.ReLU(),
            nn.Conv2d(self.out_channels, self.in_channels, 1),
        )

    def forward(self, x):
        b, c, h, w = x.size()
        # key -> [b, 1, H, W] -> [b, 1, H*W] ->  [b, H*W, 1]
        key = self.SoftMax(self.Conv_key(x).view(b, 1, -1).permute(0, 2, 1).view(b, -1, 1).contiguous())
        query = x.view(b, c, h*w)
        # [b, c, h*w] * [b, H*W, 1]
        concate_QK = torch.matmul(query, key)
        concate_QK = concate_QK.view(b, c, 1, 1).contiguous()
        value = self.Conv_value(concate_QK)
        out = x + value
        return out

if __name__ == "__main__":
    x = torch.randn((2, 1024, 24, 24))
    GCBlock = GlobalContextBlock(in_channels=1024)
    out = GCBlock(x)
    print("GCBlock output.shape:", out.shape)
    print(out)

模型测试

由于原文中没有给出详细的模型图，所以这里作者创建了一个简单的GCNet模型。
当然，GCBlock也可以加载ResNet中的任意层内，提升一些性能，作者这里偷懒了没有这么做。

Backbone选择了ResNet50（8倍下采样，使用扩张卷积）

import torch
import torch.nn as nn

class BasicBlock(nn.Module):
    expansion: int = 4
    def __init__(self, inplanes, planes, stride = 1, downsample = None, groups = 1,
        base_width = 64, dilation = 1, norm_layer = None):
        
        super(BasicBlock, self).__init__()
        if norm_layer is None:
            norm_layer = nn.BatchNorm2d
        if groups != 1 or base_width != 64:
            raise ValueError("BasicBlock only supports groups=1 and base_width=64")
        if dilation > 1:
            raise NotImplementedError("Dilation > 1 not supported in BasicBlock")
        # Both self.conv1 and self.downsample layers downsample the input when stride != 1
        self.conv1 = nn.Conv2d(inplanes, planes ,kernel_size=3, stride=stride, 
                               padding=dilation,groups=groups, bias=False,dilation=dilation)
        
        self.bn1 = norm_layer(planes)
        self.relu = nn.ReLU(inplace=True)
        self.conv2 = nn.Conv2d(planes, planes ,kernel_size=3, stride=stride, 
                               padding=dilation,groups=groups, bias=False,dilation=dilation)
        
        self.bn2 = norm_layer(planes)
        self.downsample = downsample
        self.stride = stride

    def forward(self, x):
        identity = x

        out = self.conv1(x)
        out = self.bn1(out)
        out = self.relu(out)

        out = self.conv2(out)
        out = self.bn2(out)

        if self.downsample is not None:
            identity = self.downsample(x)

        out += identity
        out = self.relu(out)

        return out


class Bottleneck(nn.Module):
    expansion = 4

    def __init__(self, inplanes, planes, stride=1, downsample= None,
        groups = 1, base_width = 64, dilation = 1, norm_layer = None,):
        super(Bottleneck, self).__init__()
        if norm_layer is None:
            norm_layer = nn.BatchNorm2d
        width = int(planes * (base_width / 64.0)) * groups
        # Both self.conv2 and self.downsample layers downsample the input when stride != 1
        self.conv1 = nn.Conv2d(inplanes, width, kernel_size=1, stride=1, bias=False)
        self.bn1 = norm_layer(width)
        self.conv2 = nn.Conv2d(width, width, kernel_size=3, stride=stride, bias=False, padding=dilation, dilation=dilation)
        self.bn2 = norm_layer(width)
        self.conv3 = nn.Conv2d(width, planes * self.expansion, kernel_size=1, stride=1, bias=False)
        self.bn3 = norm_layer(planes * self.expansion)
        self.relu = nn.ReLU(inplace=True)
        self.downsample = downsample
        self.stride = stride

    def forward(self, x):
        identity = x

        out = self.conv1(x)
        out = self.bn1(out)
        out = self.relu(out)

        out = self.conv2(out)
        out = self.bn2(out)
        out = self.relu(out)

        out = self.conv3(out)
        out = self.bn3(out)

        if self.downsample is not None:
            identity = self.downsample(x)

        out += identity
        out = self.relu(out)
        return out


class ResNet(nn.Module):
    def __init__(
        self,block, layers,num_classes = 1000, zero_init_residual = False, groups = 1,
        width_per_group = 64, replace_stride_with_dilation = None, norm_layer = None):
        super(ResNet, self).__init__()
        if norm_layer is None:
            norm_layer = nn.BatchNorm2d
        self._norm_layer = norm_layer
        self.inplanes = 64
        self.dilation = 2
        if replace_stride_with_dilation is None:
            # each element in the tuple indicates if we should replace
            # the 2x2 stride with a dilated convolution instead
            replace_stride_with_dilation = [False, False, False]
            
        if len(replace_stride_with_dilation) != 3:
            raise ValueError(
                "replace_stride_with_dilation should be None "
                f"or a 3-element tuple, got {replace_stride_with_dilation}"
            )
        self.groups = groups
        self.base_width = width_per_group
        self.conv1 = nn.Conv2d(3, self.inplanes, kernel_size=7, stride=2, padding=3, bias=False)
        self.bn1 = norm_layer(self.inplanes)
        self.relu = nn.ReLU(inplace=True)
        self.maxpool = nn.MaxPool2d(kernel_size=3, stride=2, padding=1)
        self.layer1 = self._make_layer(block, 64, layers[0])
        self.layer2 = self._make_layer(block, 128, layers[1], stride=1, dilate=replace_stride_with_dilation[0])
        self.layer3 = self._make_layer(block, 256, layers[2], stride=2, dilate=replace_stride_with_dilation[1])
        self.layer4 = self._make_layer(block, 512, layers[3], stride=1, dilate=replace_stride_with_dilation[2])
        self.avgpool = nn.AdaptiveAvgPool2d((1, 1))
        self.fc = nn.Linear(512 * block.expansion, num_classes)

        for m in self.modules():
            if isinstance(m, nn.Conv2d):
                nn.init.kaiming_normal_(m.weight, mode="fan_out", nonlinearity="relu")
            elif isinstance(m, (nn.BatchNorm2d, nn.GroupNorm)):
                nn.init.constant_(m.weight, 1)
                nn.init.constant_(m.bias, 0)

        # Zero-initialize the last BN in each residual branch,
        # so that the residual branch starts with zeros, and each residual block behaves like an identity.
        # This improves the model by 0.2~0.3% according to https://arxiv.org/abs/1706.02677
        if zero_init_residual:
            for m in self.modules():
                if isinstance(m, Bottleneck):
                    nn.init.constant_(m.bn3.weight, 0)  # type: ignore[arg-type]
                elif isinstance(m, BasicBlock):
                    nn.init.constant_(m.bn2.weight, 0)  # type: ignore[arg-type]

    def _make_layer(
        self,
        block,
        planes,
        blocks,
        stride = 1,
        dilate = False,
    ):
        norm_layer = self._norm_layer
        downsample = None
        previous_dilation = self.dilation
        if dilate:
            self.dilation *= stride
            stride = stride
            
        if stride != 1 or self.inplanes != planes * block.expansion:
            downsample = nn.Sequential(
                nn.Conv2d(self.inplanes,  planes * block.expansion, kernel_size=1, stride=stride, bias=False),
                norm_layer(planes * block.expansion))

        layers = []
        layers.append(
            block(
                self.inplanes, planes, stride, downsample, self.groups, self.base_width, previous_dilation, norm_layer
            )
        )
        self.inplanes = planes * block.expansion
        for _ in range(1, blocks):
            layers.append(
                block(
                    self.inplanes,
                    planes,
                    groups=self.groups,
                    base_width=self.base_width,
                    dilation=self.dilation,
                    norm_layer=norm_layer,
                )
            )
        return nn.Sequential(*layers)

    def _forward_impl(self, x):
        out = []
        x = self.conv1(x)
        x = self.bn1(x)
        x = self.relu(x)
        x = self.maxpool(x)
        x = self.layer1(x)
        x = self.layer2(x)
        x = self.layer3(x)
        x = self.layer4(x)

        return x
        
    def forward(self, x) :
        return self._forward_impl(x)
    def _resnet(block, layers, pretrained_path = None, **kwargs,):
        model = ResNet(block, layers, **kwargs)
        if pretrained_path is not None:
            model.load_state_dict(torch.load(pretrained_path),  strict=False)
        return model
    
    def resnet50(pretrained_path=None, **kwargs):
        return ResNet._resnet(Bottleneck, [3, 4, 6, 3],pretrained_path,**kwargs)
    
    def resnet101(pretrained_path=None, **kwargs):
        return ResNet._resnet(Bottleneck, [3, 4, 23, 3],pretrained_path,**kwargs)

GCBlock

import torch
import torch.nn as nn
import torch.nn.functional as F
class GlobalContextBlock(nn.Module):
    def __init__(self, in_channels, scale = 16):
        super(GlobalContextBlock, self).__init__()
        self.in_channels = in_channels
        self.out_channels = self.in_channels//scale

        self.Conv_key = nn.Conv2d(self.in_channels, 1, 1)
        self.SoftMax = nn.Softmax(dim=1)

        self.Conv_value = nn.Sequential(
            nn.Conv2d(self.in_channels, self.out_channels, 1),
            nn.LayerNorm([self.out_channels, 1, 1]),
            nn.ReLU(),
            nn.Conv2d(self.out_channels, self.in_channels, 1),
        )

    def forward(self, x):
        b, c, h, w = x.size()
        # key -> [b, 1, H, W] -> [b, 1, H*W] ->  [b, H*W, 1]
        key = self.SoftMax(self.Conv_key(x).view(b, 1, -1).permute(0, 2, 1).view(b, -1, 1).contiguous())
        query = x.view(b, c, h*w)
        # [b, c, h*w] * [b, H*W, 1]
        concate_QK = torch.matmul(query, key)
        concate_QK = concate_QK.view(b, c, 1, 1).contiguous()
        value = self.Conv_value(concate_QK)
        out = x + value
        return out

if __name__ == "__main__":
    x = torch.randn((2, 1024, 24, 24))
    GCBlock = GlobalContextBlock(in_channels=1024)
    out = GCBlock(x)
    print("GCBlock output.shape:", out.shape)
    print(out)

GCNet

import torch
import torch.nn as nn
import torch.nn.functional as F
class GCNet(nn.Module):
    def __init__(self, num_classes):
        super(GCNet, self).__init__()
        self.gc_block = GlobalContextBlock(in_channels=2048, scale = 16)
        self.backbone = ResNet.resnet50(replace_stride_with_dilation=[1,2,4])
        self.Conv_1 = nn.Sequential(
            nn.Conv2d(2048, 512, 3, padding=1),
            nn.BatchNorm2d(512),
            nn.ReLU(),
            nn.Upsample(scale_factor=8, mode="bilinear", align_corners=True)
        )
        self.cls_seg = nn.Conv2d(512, num_classes, 3, padding=1)

    def forward(self, x):
        """Forward function."""
        output = self.backbone(x)
        output = self.gc_block(output)
        output = self.Conv_1(output)
        output = self.cls_seg(output)
        return output

if __name__ == "__main__":
    x = torch.randn((2, 3, 224, 224))
    model = GCNet(num_classes=2)
    out = model(x)
    print("GCNet output.shape:", out.shape)

Dataset camvid

# 导入库
import os
os.environ['CUDA_VISIBLE_DEVICES'] = '0'
os.environ["KMP_DUPLICATE_LIB_OK"]="TRUE"
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.utils.data import DataLoader
from tqdm import tqdm
import warnings
warnings.filterwarnings("ignore")
import matplotlib.pyplot as plt
from PIL import Image
import numpy as np
import albumentations as A
from albumentations.pytorch.transforms import ToTensorV2
 
torch.manual_seed(17)
# 自定义数据集CamVidDataset
class CamVidDataset(torch.utils.data.Dataset):
    """CamVid Dataset. Read images, apply augmentation and preprocessing transformations.
    
    Args:
        images_dir (str): path to images folder
        masks_dir (str): path to segmentation masks folder
        class_values (list): values of classes to extract from segmentation mask
        augmentation (albumentations.Compose): data transfromation pipeline 
            (e.g. flip, scale, etc.)
        preprocessing (albumentations.Compose): data preprocessing 
            (e.g. noralization, shape manipulation, etc.)
    """
    
    def __init__(self, images_dir, masks_dir):
        self.transform = A.Compose([
            A.Resize(224, 224),
            A.HorizontalFlip(),
            A.VerticalFlip(),
            A.Normalize(),
            ToTensorV2(),
        ]) 
        self.ids = os.listdir(images_dir)
        self.images_fps = [os.path.join(images_dir, image_id) for image_id in self.ids]
        self.masks_fps = [os.path.join(masks_dir, image_id) for image_id in self.ids]
 
    
    def __getitem__(self, i):
        # read data
        image = np.array(Image.open(self.images_fps[i]).convert('RGB'))
        mask = np.array( Image.open(self.masks_fps[i]).convert('RGB'))
        image = self.transform(image=image,mask=mask)
        
        return image['image'], image['mask'][:,:,0]
        
    def __len__(self):
        return len(self.ids)
    
    
# 设置数据集路径
DATA_DIR = r'database/camvid/camvid/' # 根据自己的路径来设置
x_train_dir = os.path.join(DATA_DIR, 'train_images')
y_train_dir = os.path.join(DATA_DIR, 'train_labels')
x_valid_dir = os.path.join(DATA_DIR, 'valid_images')
y_valid_dir = os.path.join(DATA_DIR, 'valid_labels')
    
train_dataset = CamVidDataset(
    x_train_dir, 
    y_train_dir, 
)
val_dataset = CamVidDataset(
    x_valid_dir, 
    y_valid_dir, 
)
 
train_loader = DataLoader(train_dataset, batch_size=8, shuffle=True,drop_last=True)
val_loader = DataLoader(val_dataset, batch_size=8, shuffle=True,drop_last=True)

训练过程

model = GCNet(num_classes=33).cuda()

from d2l import torch as d2l
from tqdm import tqdm
import pandas as pd
import monai
# training loop 100 epochs
epochs_num = 100
# 选用SGD优化器来训练
optimizer = torch.optim.SGD(model.parameters(), lr=0.1)
schedule = monai.optimizers.LinearLR(optimizer, end_lr=0.05, num_iter=int(epochs_num*0.75))
# 损失函数选用多分类交叉熵损失函数
lossf = nn.CrossEntropyLoss(ignore_index=255)


# 训练函数
def train_ch13(net, train_iter, test_iter, loss, optimizer, num_epochs, scheduler, devices=d2l.try_all_gpus()):
    timer, num_batches = d2l.Timer(), len(train_iter)
    animator = d2l.Animator(xlabel='epoch', xlim=[1, num_epochs], ylim=[0, 1], legend=['train loss', 'train acc', 'test acc'])
    net = nn.DataParallel(net, device_ids=devices).to(devices[0])
    # 用来保存一些训练参数

    loss_list = []
    train_acc_list = []
    test_acc_list = []
    epochs_list = []
    time_list = []
    lr_list = []
    

    for epoch in range(num_epochs):
        # Sum of training loss, sum of training accuracy, no. of examples,
        # no. of predictions
        metric = d2l.Accumulator(4)
        for i, (X, labels) in enumerate(train_iter):
            timer.start()
            #l, acc = d2l.train_batch_ch13(net, features, labels.long(), loss, optimizer, devices)
            if isinstance(X, list):
                X = [x.to(devices[0]) for x in X]
            else:
                X = X.to(devices[0])
            y = labels.long().to(devices[0])

            net.train()
            optimizer.zero_grad()
            pred = net(X)
            l = loss(pred, y)
            l.sum().backward()
            optimizer.step()
            l = l.sum()
            acc = d2l.accuracy(pred, y)
            metric.add(l, acc, labels.shape[0], labels.numel())
            timer.stop()
            if (i + 1) % (num_batches // 5) == 0 or i == num_batches - 1:
                animator.add(epoch + (i + 1) / num_batches,(metric[0] / metric[2], metric[1] / metric[3], None))
                
        if epoch < 75:
            schedule.step()
        test_acc = d2l.evaluate_accuracy_gpu(net, test_iter)
        animator.add(epoch + 1, (None, None, test_acc))

        print(f"epoch {epoch+1}/{epochs_num} --- loss {metric[0] / metric[2]:.3f} --- train acc {metric[1] / metric[3]:.3f} --- test acc {test_acc:.3f} --- lr {optimizer.state_dict()['param_groups'][0]['lr']} --- cost time {timer.sum()}")
        
        #---------保存训练数据---------------
        df = pd.DataFrame()
        loss_list.append(metric[0] / metric[2])
        train_acc_list.append(metric[1] / metric[3])
        test_acc_list.append(test_acc)
        epochs_list.append(epoch+1)
        time_list.append(timer.sum())
        lr_list.append(optimizer.state_dict()['param_groups'][0]['lr'])
        
        df['epoch'] = epochs_list
        df['loss'] = loss_list
        df['train_acc'] = train_acc_list
        df['test_acc'] = test_acc_list
        df["lr"] = lr_list
        df['time'] = time_list
        
        df.to_excel("savefile/GCNet_camvid.xlsx")
        #----------------保存模型------------------- 
        if np.mod(epoch+1, 5) == 0:
            torch.save(model.state_dict(), f'checkpoints/GCNet_{epoch+1}.pth')

    # 保存下最后的model
    torch.save(model.state_dict(), f'checkpoints/GCNet_last.pth')
    
train_ch13(model, train_loader, val_loader, lossf, optimizer, epochs_num, scheduler=schedule)

训练结果