语义分割系列17-Non-Local Net（pytorch实现）

Non-Local Net：《Non-Local Neural Networks》

发布于2018CVPR。

---

引文 

什么是Local和Non Local?

在语义分割论文中，最多被提到的词恐怕就是receptive field了，也就是感受野。也就是，天下模型苦receptive field久矣。如之前写的文章中所提到的，大家在增加感受野的方式上基本大同小异，无非就是扩大卷积核、用扩张卷积（空洞卷积）、多叠几个卷积层，当然，这些操作都有一定的扩张感受野能力，但是缺点也是显而易见的，包括不限于增加计算量、损失邻域信息等等。所以，可以说，卷积操作是Local的，因为只有卷积核覆盖的那一个局部的信息有交互，再远了就没了。

而Attention is All you need横空出世后，大家发现，原来Attention机制（尤其是指空间注意力）可以无视距离，来建立两个位置之间的信息交互（就像峡谷里死歌开大，跑都跑不掉那种），那这就是Non-Local的操作。

当然，全连接其实也是Non Local的，但是真的全部连接的话，参数量也是大的惊人。

为了解决这个感受野受限的问题，大家就开始对Attention机制开始各种魔改啊。其中，我认为就包括了Non-Local的操作（至少在我浅薄的知识体系中，我认为Non-Local无非是一个Attention的变体）。

论文思想

Non-Local的思想一张图就能概括完，乍一看，这三个输入Θ、Φ、g，不就对应着Attention机制里面的qkv么。确实结构也很像。

不妨回顾一下之前写过DANet中的PAM结构

啊这。一模一样有没有。当然DANet是2019年的工作，看到NonLocal还得喊一声大哥。

当然，这里的Non-Local有一个比较不同的地方，我们都知道，attention需要计算一个相似度，而这里NonLocal选择了几个相似度计算的方法，包括Gaussian版本、embedded_gaussian、concatenation还有点积版本的，很显然，DANet中的PAM使用的是点积版本的。在NonLocal论文中使用的是Gaussian版本的。

那我们再回到Non-Local结构中来，把图在下面重新放一下。

NL结构用一个公式就能概括，和Attention类似。

对于每一个点，也可以写成：

其中i, j是两个点的位置。新计算出的y即使i位置的新值或者叫Attention激活值。g（）函数比较简单，线性嵌入即可，用一个卷积核来实现，而这个f（）函数就是作者介绍的几种相似度的计算函数，包括有：

Gaussian

Embedded Gaussian

Dot product

Concatenation

 

小结

Non-Local机制类似于空间Attention，被引入来建立任意距离的位置的链接。后续很多工作也在Non-Local机制上做了一些改动。

模型复现

由于本文注重于二维的图像分割，所以只实现了2D的Non-Local模块，并瞎写了一个模型来实现，瞎写的模型大抵类似于DeepLabv3+结构。

backbone-resnet50（8倍下采样）

import torch
import torch.nn as nn
import torch.nn.functional as F
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 ,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 ,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):
        x = self.conv1(x)
        x = self.bn1(x)
        x = self.relu(x)
        x = self.maxpool(x)
        x = self.layer1(x)
        x = self.layer2(x)
        x1 = x
        x = self.layer3(x)
        
        x = self.layer4(x)
        return x1, 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)

Non-Local

import torch
import torch.nn as nn

class NonLocal2d(nn.Module):
    def __init__(self, in_channels, reduction=2, use_scale=True, sub_sample=False, mode='embedded_gaussian'):
        super(NonLocal2d, self).__init__()
        self.in_channels = in_channels
        self.reduction = reduction
        self.use_scale = use_scale
        self.inter_channels = max(in_channels // reduction, 1)
        self.mode = mode

        if mode not in [
                'gaussian', 'embedded_gaussian', 'dot_product', 'concatenation'
        ]:
            raise ValueError("Mode should be in 'gaussian', 'concatenation', "
                             f"'embedded_gaussian' or 'dot_product', but got "
                             f'{mode} instead.')

        self.g = nn.Conv2d(
            self.in_channels,
            self.inter_channels,
            kernel_size=1,)
        self.conv_out = nn.Conv2d(
            self.inter_channels,
            self.in_channels,
            kernel_size=1,)

        if self.mode != 'gaussian':
            self.theta = nn.Conv2d(
                self.in_channels,
                self.inter_channels,
                kernel_size=1,)
            self.phi = nn.Conv2d(
                self.in_channels,
                self.inter_channels,
                kernel_size=1,)

        if self.mode == 'concatenation':
            self.concat_project = nn.Sequential(
                nn.Conv2d(
                    self.inter_channels * 2,
                    1,
                    kernel_size=1,
                    stride=1,
                    padding=0,
                    bias=False,
                ),
                nn.ReLU())
        self.sub_sample = sub_sample
        if sub_sample:
            max_pool_layer = nn.MaxPool2d(kernel_size=(2, 2))
            self.g = nn.Sequential(self.g, max_pool_layer)
            if self.mode != 'gaussian':
                self.phi = nn.Sequential(self.phi, max_pool_layer)
            else:
                self.phi = max_pool_layer

        self.init_weights()




    def init_weights(self, std=0.01, zeros_init=True):
        if self.mode != 'gaussian':
            for m in [self.g, self.theta, self.phi]:
                nn.init.normal_(m.weight.data, std=std)
        else:
            nn.init.normal_(self.g.weight.data, std=std)
            
        if zeros_init:
            nn.init.normal_(self.conv_out.weight.data, 0)
        else:
            nn.init.normal_(self.conv_out.weight.data, std=std)



    def gaussian(self, theta_x, phi_x):
        # NonLocal2d pairwise_weight: [N, HxW, HxW]
        pairwise_weight = torch.matmul(theta_x, phi_x)
        pairwise_weight = pairwise_weight.softmax(dim=-1)
        return pairwise_weight

    def embedded_gaussian(self, theta_x, phi_x):
        # NonLocal2d pairwise_weight: [N, HxW, HxW]
        pairwise_weight = torch.matmul(theta_x, phi_x)
        if self.use_scale:
            # theta_x.shape[-1] is `self.inter_channels`
            pairwise_weight /= theta_x.shape[-1]**0.5
        pairwise_weight = pairwise_weight.softmax(dim=-1)
        return pairwise_weight

    def dot_product(self, theta_x, phi_x):
        # NonLocal2d pairwise_weight: [N, HxW, HxW]
        pairwise_weight = torch.matmul(theta_x, phi_x)
        pairwise_weight /= pairwise_weight.shape[-1]
        return pairwise_weight

    def concatenation(self, theta_x, phi_x):
        # NonLocal2d pairwise_weight: [N, HxW, HxW]
        h = theta_x.size(2)
        w = phi_x.size(3)
        theta_x = theta_x.repeat(1, 1, 1, w)
        phi_x = phi_x.repeat(1, 1, h, 1)

        concat_feature = torch.cat([theta_x, phi_x], dim=1)
        pairwise_weight = self.concat_project(concat_feature)
        n, _, h, w = pairwise_weight.size()
        pairwise_weight = pairwise_weight.view(n, h, w)
        pairwise_weight /= pairwise_weight.shape[-1]

        return pairwise_weight

    def forward(self, x):
        # NonLocal2d x: [N, C, H, W]
        n = x.size(0)
        # NonLocal2d g_x: [N, HxW, C]
        g_x = self.g(x).view(n, self.inter_channels, -1)
        g_x = g_x.permute(0, 2, 1)
        # NonLocal2d theta_x: [N, HxW, C], phi_x: [N, C, HxW]
        if self.mode == 'gaussian':
            theta_x = x.view(n, self.in_channels, -1)
            theta_x = theta_x.permute(0, 2, 1)
            if self.sub_sample:
                phi_x = self.phi(x).view(n, self.in_channels, -1)
            else:
                phi_x = x.view(n, self.in_channels, -1)
        elif self.mode == 'concatenation':
            theta_x = self.theta(x).view(n, self.inter_channels, -1, 1)
            phi_x = self.phi(x).view(n, self.inter_channels, 1, -1)
        else:
            theta_x = self.theta(x).view(n, self.inter_channels, -1)
            theta_x = theta_x.permute(0, 2, 1)
            phi_x = self.phi(x).view(n, self.inter_channels, -1)

        pairwise_func = getattr(self, self.mode)
        # NonLocal1d pairwise_weight: [N, H, H]
        # NonLocal2d pairwise_weight: [N, HxW, HxW]
        # NonLocal3d pairwise_weight: [N, TxHxW, TxHxW]
        pairwise_weight = pairwise_func(theta_x, phi_x)

        # NonLocal2d y: [N, HxW, C]
        y = torch.matmul(pairwise_weight, g_x)
        # NonLocal2d y: [N, C, H, W]
        y = y.permute(0, 2, 1).contiguous().reshape(n, self.inter_channels,
                                                    *x.size()[2:])
        output = x + self.conv_out(y)

        return output

class NLHead(nn.Module):
    def __init__(self, reduction=2, use_scale=True, mode='embedded_gaussian', channels=2048):
        super(NLHead, self).__init__()
        self.reduction = reduction
        self.use_scale = use_scale
        self.mode = mode
        self.channels = channels
        self.nl_block = NonLocal2d(
            in_channels=self.channels,
            reduction=self.reduction,
            use_scale=self.use_scale,
            mode=self.mode)
        self.conv1 = nn.Sequential(
            nn.Conv2d(self.channels, self.channels, 3, padding=1),
        )
        self.conv2 = nn.Sequential(
            nn.Conv2d(self.channels, self.channels, 3, padding=1),
        )
        self.cat = False
        
    def forward(self, inputs):
        """Forward function."""
        x = inputs
        output = self.conv1(x)
        output = self.nl_block(output)
        output = self.conv2(output)
        if self.cat:
            output = torch.cat([x, output], dim=1)
        return output

Model

class NonLocalNet(nn.Module):
    def __init__(self, num_classes):
        super(NonLocalNet, self).__init__()
        self.num_classes = num_classes
        self.resnet = ResNet.resnet50(replace_stride_with_dilation=[1,2,4])
        self.NLHead = NLHead()
        self.upSample = nn.Sequential(
            nn.Conv2d(2048, 512, 3, padding=1),
            nn.BatchNorm2d(512),
            nn.ReLU(),
            nn.Upsample(scale_factor=2., mode="bilinear", align_corners=True),
        )
        
        self.cls_seg = nn.Sequential(
            nn.Conv2d(1024, 256, 3, padding=1),
            nn.BatchNorm2d(256),
            nn.ReLU(),
            nn.Upsample(scale_factor=4., mode="bilinear", align_corners=True),
            nn.Conv2d(256, self.num_classes, 3, padding=1),  
        )
        
    def forward(self, x):

        x_4, x_8 = self.resnet(x)
        x_8 = self.NLHead(x_8)
        x_8 = self.upSample(x_8)
        x = torch.cat([x_8, x_4], 1)
        x = self.cls_seg(x)
        return x

Dataset-Camvid

# 导入库
import os
os.environ['CUDA_VISIBLE_DEVICES'] = '0'
 
import torch
import torch.nn as nn
import torch.optim as optim
import torch.nn.functional as F
from torch import optim
from torch.utils.data import Dataset, DataLoader, random_split
from tqdm import tqdm
import warnings
warnings.filterwarnings("ignore")
import os.path as osp
from PIL import Image
import numpy as np
import albumentations as A
from albumentations.pytorch.transforms import ToTensorV2
import matplotlib.pyplot as plt

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)

Train

model = NonLocalNet(num_classes=33).cuda()
#载入预训练模型
#model.load_state_dict(torch.load(r"checkpoints/Unet++_25.pth"),strict=False)

from d2l import torch as d2l
from tqdm import tqdm
import pandas as pd
#损失函数选用多分类交叉熵损失函数
lossf = nn.CrossEntropyLoss(ignore_index=255)
#选用adam优化器来训练
optimizer = optim.SGD(model.parameters(), lr=0.1)
scheduler = torch.optim.lr_scheduler.StepLR(optimizer, step_size=50, gamma=0.5, last_epoch=-1)

#训练50轮
epochs_num = 100
def train_ch13(net, train_iter, test_iter, loss, trainer, 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])
    
    loss_list = []
    train_acc_list = []
    test_acc_list = []
    epochs_list = []
    time_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, (features, labels) in enumerate(train_iter):
            timer.start()
            l, acc = d2l.train_batch_ch13(
                net, features, labels.long(), loss, trainer, devices)
            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))
        test_acc = d2l.evaluate_accuracy_gpu(net, test_iter)
        animator.add(epoch + 1, (None, None, test_acc))
        scheduler.step()
        print(f"epoch {epoch+1} --- loss {metric[0] / metric[2]:.3f} ---  train acc {metric[1] / metric[3]:.3f} --- test acc {test_acc:.3f} --- 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())
        
        df['epoch'] = epochs_list
        df['loss'] = loss_list
        df['train_acc'] = train_acc_list
        df['test_acc'] = test_acc_list
        df['time'] = time_list
        df.to_excel("savefile/NonLocalNet_camvid.xlsx")
        #----------------保存模型-------------------
        if np.mod(epoch+1, 5) == 0:
            torch.save(model.state_dict(), f'checkpoints/NonLocalNet_{epoch+1}.pth')
train_ch13(model, train_loader, val_loader, lossf, optimizer, epochs_num,scheduler)

结果