【人脸识别】arcface-pytorch代码解析

参考：

https://github.com/ronghuaiyang/arcface-pytorch

1.代码结构

目录结构如下：

配置信息：包含许多训练或者测试的一些信息配置，比如backbone选用的模型结构等；

数据读取：里面是一个pytorch的Dataloder，我们可以自定义，其中__getitem__用于迭代时输出图像与label数据对；

模型：这里面含有backbone结构，loss损失函数等结构；

脚本：这里面放了一些其他的用于后续处理的脚本。

test：测试脚本

train：训练脚本

2.关键脚本解析

2.1 config/config.py文件

这里是使用一个类设置初始变量配置信息。在调用的使用只需要实例化这个类，然后用这个对象名称加上.env（里面的参数名，就可以调用了）

class Config(object):
    env = 'default'
    backbone = 'resnet18'
    classify = 'softmax'
    num_classes = 13938
    metric = 'arc_margin'
    easy_margin = False
    use_se = False
    loss = 'focal_loss'

    display = False
    finetune = False

    train_root = '/data/Datasets/webface/CASIA-maxpy-clean-crop-144/'
    train_list = '/data/Datasets/webface/train_data_13938.txt'
    val_list = '/data/Datasets/webface/val_data_13938.txt'

    test_root = '/data1/Datasets/anti-spoofing/test/data_align_256'
    test_list = 'test.txt'

    lfw_root = '/data/Datasets/lfw/lfw-align-128'
    lfw_test_list = '/data/Datasets/lfw/lfw_test_pair.txt'

    checkpoints_path = 'checkpoints'
    load_model_path = 'models/resnet18.pth'
    test_model_path = 'checkpoints/resnet18_110.pth'
    save_interval = 10

    train_batch_size = 16  # batch size
    test_batch_size = 60

    input_shape = (1, 128, 128)

    optimizer = 'sgd'

    use_gpu = True  # use GPU or not
    gpu_id = '0, 1'
    num_workers = 4  # how many workers for loading data
    print_freq = 100  # print info every N batch

    debug_file = '/tmp/debug'  # if os.path.exists(debug_file): enter ipdb
    result_file = 'result.csv'

    max_epoch = 50
    lr = 1e-1  # initial learning rate
    lr_step = 10
    lr_decay = 0.95  # when val_loss increase, lr = lr*lr_decay
    weight_decay = 5e-4

2.2 data/dataset.py文件

参考此处：【从DataLoader到Model（一）】三步法写自定义Torch的DataLoader - 知乎

我们会发现下面的dataloader是一个标准的dataloader格式，包含__init__,__getitem__,__len__，这几个方法。其中__init__中定义了图像从文件目录中读取到一个list中，以及train模式下与其他模式下要进行的一些tranforms的一些数据增强操作。__getitem__会读取图像，并进行初始化过程中的一些transforms操作，并返回图像与labels。__len__返回图像的总数目。通过main函数中，前面几行就可以按批次迭代地调用数据了。

import os
from PIL import Image
import torch
from torch.utils import data
import numpy as np
from torchvision import transforms as T
import torchvision
import cv2
import sys


class Dataset(data.Dataset):

    def __init__(self, root, data_list_file, phase='train', input_shape=(1, 128, 128)):
        self.phase = phase
        self.input_shape = input_shape

        with open(os.path.join(data_list_file), 'r') as fd:
            imgs = fd.readlines()

        imgs = [os.path.join(root, img[:-1]) for img in imgs]
        self.imgs = np.random.permutation(imgs)

        # normalize = T.Normalize(mean=[0.5, 0.5, 0.5],
        #                         std=[0.5, 0.5, 0.5])

        normalize = T.Normalize(mean=[0.5], std=[0.5])

        if self.phase == 'train':
            self.transforms = T.Compose([
                T.RandomCrop(self.input_shape[1:]),
                T.RandomHorizontalFlip(),
                T.ToTensor(),
                normalize
            ])
        else:
            self.transforms = T.Compose([
                T.CenterCrop(self.input_shape[1:]),
                T.ToTensor(),
                normalize
            ])

    def __getitem__(self, index):
        sample = self.imgs[index]
        splits = sample.split()
        img_path = splits[0]
        data = Image.open(img_path)
        data = data.convert('L')
        data = self.transforms(data)
        label = np.int32(splits[1])
        return data.float(), label

    def __len__(self):
        return len(self.imgs)


if __name__ == '__main__':
    dataset = Dataset(root='/data/Datasets/fv/dataset_v1.1/dataset_mix_aligned_v1.1',
                      data_list_file='/data/Datasets/fv/dataset_v1.1/mix_20w.txt',
                      phase='test',
                      input_shape=(1, 128, 128))

    trainloader = data.DataLoader(dataset, batch_size=10)
    for i, (data, label) in enumerate(trainloader):
        # imgs, labels = data
        # print imgs.numpy().shape
        # print data.cpu().numpy()
        # if i == 0:
        img = torchvision.utils.make_grid(data).numpy()
        # print img.shape
        # print label.shape
        # chw -> hwc
        img = np.transpose(img, (1, 2, 0))
        # img *= np.array([0.229, 0.224, 0.225])
        # img += np.array([0.485, 0.456, 0.406])
        img += np.array([1, 1, 1])
        img *= 127.5
        img = img.astype(np.uint8)
        img = img[:, :, [2, 1, 0]]

        cv2.imshow('img', img)
        cv2.waitKey()
        # break
        # dst.decode_segmap(labels.numpy()[0], plot=True)

2.3 models/resnet.py

下面的结构应该是在resnet基础上修改的，包含resnetface。

# -*- coding: utf-8 -*-
"""
Created on 18-5-21 下午5:26

@author: ronghuaiyang
"""
import torch
import torch.nn as nn
import math
import torch.utils.model_zoo as model_zoo
import torch.nn.utils.weight_norm as weight_norm
import torch.nn.functional as F


# __all__ = ['ResNet', 'resnet18', 'resnet34', 'resnet50', 'resnet101',
#            'resnet152']


model_urls = {
    'resnet18': 'https://download.pytorch.org/models/resnet18-5c106cde.pth',
    'resnet34': 'https://download.pytorch.org/models/resnet34-333f7ec4.pth',
    'resnet50': 'https://download.pytorch.org/models/resnet50-19c8e357.pth',
    'resnet101': 'https://download.pytorch.org/models/resnet101-5d3b4d8f.pth',
    'resnet152': 'https://download.pytorch.org/models/resnet152-b121ed2d.pth',
}


def conv3x3(in_planes, out_planes, stride=1):
    """3x3 convolution with padding"""
    return nn.Conv2d(in_planes, out_planes, kernel_size=3, stride=stride,
                     padding=1, bias=False)


class BasicBlock(nn.Module):
    expansion = 1

    def __init__(self, inplanes, planes, stride=1, downsample=None):
        super(BasicBlock, self).__init__()
        self.conv1 = conv3x3(inplanes, planes, stride)
        self.bn1 = nn.BatchNorm2d(planes)
        self.relu = nn.ReLU(inplace=True)
        self.conv2 = conv3x3(planes, planes)
        self.bn2 = nn.BatchNorm2d(planes)
        self.downsample = downsample
        self.stride = stride

    def forward(self, x):
        residual = 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:
            residual = self.downsample(x)

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

        return out


class IRBlock(nn.Module):
    expansion = 1

    def __init__(self, inplanes, planes, stride=1, downsample=None, use_se=True):
        super(IRBlock, self).__init__()
        self.bn0 = nn.BatchNorm2d(inplanes)
        self.conv1 = conv3x3(inplanes, inplanes)
        self.bn1 = nn.BatchNorm2d(inplanes)
        self.prelu = nn.PReLU()
        self.conv2 = conv3x3(inplanes, planes, stride)
        self.bn2 = nn.BatchNorm2d(planes)
        self.downsample = downsample
        self.stride = stride
        self.use_se = use_se
        if self.use_se:
            self.se = SEBlock(planes)

    def forward(self, x):
        residual = x
        out = self.bn0(x)
        out = self.conv1(out)
        out = self.bn1(out)
        out = self.prelu(out)

        out = self.conv2(out)
        out = self.bn2(out)
        if self.use_se:
            out = self.se(out)

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

        out += residual
        out = self.prelu(out)

        return out


class Bottleneck(nn.Module):
    expansion = 4

    def __init__(self, inplanes, planes, stride=1, downsample=None):
        super(Bottleneck, self).__init__()
        self.conv1 = nn.Conv2d(inplanes, planes, kernel_size=1, bias=False)
        self.bn1 = nn.BatchNorm2d(planes)
        self.conv2 = nn.Conv2d(planes, planes, kernel_size=3, stride=stride,
                               padding=1, bias=False)
        self.bn2 = nn.BatchNorm2d(planes)
        self.conv3 = nn.Conv2d(planes, planes * self.expansion, kernel_size=1, bias=False)
        self.bn3 = nn.BatchNorm2d(planes * self.expansion)
        self.relu = nn.ReLU(inplace=True)
        self.downsample = downsample
        self.stride = stride

    def forward(self, x):
        residual = 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:
            residual = self.downsample(x)

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

        return out


class SEBlock(nn.Module):
    def __init__(self, channel, reduction=16):
        super(SEBlock, self).__init__()
        self.avg_pool = nn.AdaptiveAvgPool2d(1)
        self.fc = nn.Sequential(
                nn.Linear(channel, channel // reduction),
                nn.PReLU(),
                nn.Linear(channel // reduction, channel),
                nn.Sigmoid()
        )

    def forward(self, x):
        b, c, _, _ = x.size()
        y = self.avg_pool(x).view(b, c)
        y = self.fc(y).view(b, c, 1, 1)
        return x * y


class ResNetFace(nn.Module):
    def __init__(self, block, layers, use_se=True):
        self.inplanes = 64
        self.use_se = use_se
        super(ResNetFace, self).__init__()
        self.conv1 = nn.Conv2d(1, 64, kernel_size=3, padding=1, bias=False)
        self.bn1 = nn.BatchNorm2d(64)
        self.prelu = nn.PReLU()
        self.maxpool = nn.MaxPool2d(kernel_size=2, stride=2)
        self.layer1 = self._make_layer(block, 64, layers[0])
        self.layer2 = self._make_layer(block, 128, layers[1], stride=2)
        self.layer3 = self._make_layer(block, 256, layers[2], stride=2)
        self.layer4 = self._make_layer(block, 512, layers[3], stride=2)
        self.bn4 = nn.BatchNorm2d(512)
        self.dropout = nn.Dropout()
        self.fc5 = nn.Linear(512 * 8 * 8, 512)
        self.bn5 = nn.BatchNorm1d(512)

        for m in self.modules():
            if isinstance(m, nn.Conv2d):
                nn.init.xavier_normal_(m.weight)
            elif isinstance(m, nn.BatchNorm2d) or isinstance(m, nn.BatchNorm1d):
                nn.init.constant_(m.weight, 1)
                nn.init.constant_(m.bias, 0)
            elif isinstance(m, nn.Linear):
                nn.init.xavier_normal_(m.weight)
                nn.init.constant_(m.bias, 0)

    def _make_layer(self, block, planes, blocks, stride=1):
        downsample = None
        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),
                nn.BatchNorm2d(planes * block.expansion),
            )
        layers = []
        layers.append(block(self.inplanes, planes, stride, downsample, use_se=self.use_se))
        self.inplanes = planes
        for i in range(1, blocks):
            layers.append(block(self.inplanes, planes, use_se=self.use_se))

        return nn.Sequential(*layers)

    def forward(self, x):
        x = self.conv1(x)
        x = self.bn1(x)
        x = self.prelu(x)
        x = self.maxpool(x)

        x = self.layer1(x)
        x = self.layer2(x)
        x = self.layer3(x)
        x = self.layer4(x)
        x = self.bn4(x)
        x = self.dropout(x)
        x = x.view(x.size(0), -1)
        x = self.fc5(x)
        x = self.bn5(x)

        return x


class ResNet(nn.Module):

    def __init__(self, block, layers):
        self.inplanes = 64
        super(ResNet, self).__init__()
        # self.conv1 = nn.Conv2d(3, 64, kernel_size=7, stride=2, padding=3,
        #                        bias=False)
        self.conv1 = nn.Conv2d(1, 64, kernel_size=3, stride=1, padding=1,
                               bias=False)
        self.bn1 = nn.BatchNorm2d(64)
        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], stride=2)
        self.layer2 = self._make_layer(block, 128, layers[1], stride=2)
        self.layer3 = self._make_layer(block, 256, layers[2], stride=2)
        self.layer4 = self._make_layer(block, 512, layers[3], stride=2)
        # self.avgpool = nn.AvgPool2d(8, stride=1)
        # self.fc = nn.Linear(512 * block.expansion, num_classes)
        self.fc5 = nn.Linear(512 * 8 * 8, 512)

        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.init.constant_(m.weight, 1)
                nn.init.constant_(m.bias, 0)

    def _make_layer(self, block, planes, blocks, stride=1):
        downsample = None
        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),
                nn.BatchNorm2d(planes * block.expansion),
            )

        layers = []
        layers.append(block(self.inplanes, planes, stride, downsample))
        self.inplanes = planes * block.expansion
        for i in range(1, blocks):
            layers.append(block(self.inplanes, planes))

        return nn.Sequential(*layers)

    def forward(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)
        x = self.layer3(x)
        x = self.layer4(x)
        # x = nn.AvgPool2d(kernel_size=x.size()[2:])(x)
        # x = self.avgpool(x)
        x = x.view(x.size(0), -1)
        x = self.fc5(x)

        return x


def resnet18(pretrained=False, **kwargs):
    """Constructs a ResNet-18 model.
    Args:
        pretrained (bool): If True, returns a model pre-trained on ImageNet
    """
    model = ResNet(BasicBlock, [2, 2, 2, 2], **kwargs)
    if pretrained:
        model.load_state_dict(model_zoo.load_url(model_urls['resnet18']))
    return model


def resnet34(pretrained=False, **kwargs):
    """Constructs a ResNet-34 model.
    Args:
        pretrained (bool): If True, returns a model pre-trained on ImageNet
    """
    model = ResNet(BasicBlock, [3, 4, 6, 3], **kwargs)
    if pretrained:
        model.load_state_dict(model_zoo.load_url(model_urls['resnet34']))
    return model


def resnet50(pretrained=False, **kwargs):
    """Constructs a ResNet-50 model.
    Args:
        pretrained (bool): If True, returns a model pre-trained on ImageNet
    """
    model = ResNet(Bottleneck, [3, 4, 6, 3], **kwargs)
    if pretrained:
        model.load_state_dict(model_zoo.load_url(model_urls['resnet50']))
    return model


def resnet101(pretrained=False, **kwargs):
    """Constructs a ResNet-101 model.
    Args:
        pretrained (bool): If True, returns a model pre-trained on ImageNet
    """
    model = ResNet(Bottleneck, [3, 4, 23, 3], **kwargs)
    if pretrained:
        model.load_state_dict(model_zoo.load_url(model_urls['resnet101']))
    return model


def resnet152(pretrained=False, **kwargs):
    """Constructs a ResNet-152 model.
    Args:
        pretrained (bool): If True, returns a model pre-trained on ImageNet
    """
    model = ResNet(Bottleneck, [3, 8, 36, 3], **kwargs)
    if pretrained:
        model.load_state_dict(model_zoo.load_url(model_urls['resnet152']))
    return model


def resnet_face18(use_se=True, **kwargs):
    model = ResNetFace(IRBlock, [2, 2, 2, 2], use_se=use_se, **kwargs)
    return model

2.4 models/metrics.py

这里应该是度量学习的一些指标，最后一层的特征表示成什么样，具体后续补充。

from __future__ import print_function
from __future__ import division
import torch
import torch.nn as nn
import torch.nn.functional as F
from torch.nn import Parameter
import math


class ArcMarginProduct(nn.Module):
    r"""Implement of large margin arc distance: :
        Args:
            in_features: size of each input sample
            out_features: size of each output sample
            s: norm of input feature
            m: margin

            cos(theta + m)
        """
    def __init__(self, in_features, out_features, s=30.0, m=0.50, easy_margin=False):
        super(ArcMarginProduct, self).__init__()
        self.in_features = in_features
        self.out_features = out_features
        self.s = s
        self.m = m
        self.weight = Parameter(torch.FloatTensor(out_features, in_features))
        nn.init.xavier_uniform_(self.weight)

        self.easy_margin = easy_margin
        self.cos_m = math.cos(m)
        self.sin_m = math.sin(m)
        self.th = math.cos(math.pi - m)
        self.mm = math.sin(math.pi - m) * m

    def forward(self, input, label):
        # --------------------------- cos(theta) & phi(theta) ---------------------------
        cosine = F.linear(F.normalize(input), F.normalize(self.weight))
        sine = torch.sqrt((1.0 - torch.pow(cosine, 2)).clamp(0, 1))
        phi = cosine * self.cos_m - sine * self.sin_m
        if self.easy_margin:
            phi = torch.where(cosine > 0, phi, cosine)
        else:
            phi = torch.where(cosine > self.th, phi, cosine - self.mm)
        # --------------------------- convert label to one-hot ---------------------------
        # one_hot = torch.zeros(cosine.size(), requires_grad=True, device='cuda')
        one_hot = torch.zeros(cosine.size(), device='cuda')
        one_hot.scatter_(1, label.view(-1, 1).long(), 1)
        # -------------torch.where(out_i = {x_i if condition_i else y_i) -------------
        output = (one_hot * phi) + ((1.0 - one_hot) * cosine)  # you can use torch.where if your torch.__version__ is 0.4
        output *= self.s
        # print(output)

        return output


class AddMarginProduct(nn.Module):
    r"""Implement of large margin cosine distance: :
    Args:
        in_features: size of each input sample
        out_features: size of each output sample
        s: norm of input feature
        m: margin
        cos(theta) - m
    """

    def __init__(self, in_features, out_features, s=30.0, m=0.40):
        super(AddMarginProduct, self).__init__()
        self.in_features = in_features
        self.out_features = out_features
        self.s = s
        self.m = m
        self.weight = Parameter(torch.FloatTensor(out_features, in_features))
        nn.init.xavier_uniform_(self.weight)

    def forward(self, input, label):
        # --------------------------- cos(theta) & phi(theta) ---------------------------
        cosine = F.linear(F.normalize(input), F.normalize(self.weight))
        phi = cosine - self.m
        # --------------------------- convert label to one-hot ---------------------------
        one_hot = torch.zeros(cosine.size(), device='cuda')
        # one_hot = one_hot.cuda() if cosine.is_cuda else one_hot
        one_hot.scatter_(1, label.view(-1, 1).long(), 1)
        # -------------torch.where(out_i = {x_i if condition_i else y_i) -------------
        output = (one_hot * phi) + ((1.0 - one_hot) * cosine)  # you can use torch.where if your torch.__version__ is 0.4
        output *= self.s
        # print(output)

        return output

    def __repr__(self):
        return self.__class__.__name__ + '(' \
               + 'in_features=' + str(self.in_features) \
               + ', out_features=' + str(self.out_features) \
               + ', s=' + str(self.s) \
               + ', m=' + str(self.m) + ')'


class SphereProduct(nn.Module):
    r"""Implement of large margin cosine distance: :
    Args:
        in_features: size of each input sample
        out_features: size of each output sample
        m: margin
        cos(m*theta)
    """
    def __init__(self, in_features, out_features, m=4):
        super(SphereProduct, self).__init__()
        self.in_features = in_features
        self.out_features = out_features
        self.m = m
        self.base = 1000.0
        self.gamma = 0.12
        self.power = 1
        self.LambdaMin = 5.0
        self.iter = 0
        self.weight = Parameter(torch.FloatTensor(out_features, in_features))
        nn.init.xavier_uniform(self.weight)

        # duplication formula
        self.mlambda = [
            lambda x: x ** 0,
            lambda x: x ** 1,
            lambda x: 2 * x ** 2 - 1,
            lambda x: 4 * x ** 3 - 3 * x,
            lambda x: 8 * x ** 4 - 8 * x ** 2 + 1,
            lambda x: 16 * x ** 5 - 20 * x ** 3 + 5 * x
        ]

    def forward(self, input, label):
        # lambda = max(lambda_min,base*(1+gamma*iteration)^(-power))
        self.iter += 1
        self.lamb = max(self.LambdaMin, self.base * (1 + self.gamma * self.iter) ** (-1 * self.power))

        # --------------------------- cos(theta) & phi(theta) ---------------------------
        cos_theta = F.linear(F.normalize(input), F.normalize(self.weight))
        cos_theta = cos_theta.clamp(-1, 1)
        cos_m_theta = self.mlambda[self.m](cos_theta)
        theta = cos_theta.data.acos()
        k = (self.m * theta / 3.14159265).floor()
        phi_theta = ((-1.0) ** k) * cos_m_theta - 2 * k
        NormOfFeature = torch.norm(input, 2, 1)

        # --------------------------- convert label to one-hot ---------------------------
        one_hot = torch.zeros(cos_theta.size())
        one_hot = one_hot.cuda() if cos_theta.is_cuda else one_hot
        one_hot.scatter_(1, label.view(-1, 1), 1)

        # --------------------------- Calculate output ---------------------------
        output = (one_hot * (phi_theta - cos_theta) / (1 + self.lamb)) + cos_theta
        output *= NormOfFeature.view(-1, 1)

        return output

    def __repr__(self):
        return self.__class__.__name__ + '(' \
               + 'in_features=' + str(self.in_features) \
               + ', out_features=' + str(self.out_features) \
               + ', m=' + str(self.m) + ')'

2.5 utils/view_model.py

大概是输出模型每层的参数量的一个函数。

import torch
from torch.autograd import Variable
import torch.nn as nn
from graphviz import Digraph


__all__ = ['view_model']


class CNN(nn.Module):
    def __init__(self):
        super(CNN, self).__init__()
        self.conv1 = nn.Sequential(
            nn.Conv2d(in_channels=1, out_channels=16, kernel_size=5, stride=1, padding=2),
            nn.ReLU(),
            nn.MaxPool2d(kernel_size=2)
        )
        self.conv2 = nn.Sequential(
            nn.Conv2d(in_channels=16, out_channels=32, kernel_size=5, stride=1, padding=2),
            nn.ReLU(),
            nn.MaxPool2d(kernel_size=2)
        )
        self.out = nn.Linear(32*7*7, 10)

    def forward(self, x):
        x = self.conv1(x)
        x = self.conv2(x)
        x = x.view(x.size(0), -1)  # (batch, 32*7*7)
        out = self.out(x)
        return out


def make_dot(var, params=None):
    """ Produces Graphviz representation of PyTorch autograd graph
    Blue nodes are the Variables that require grad, orange are Tensors
    saved for backward in torch.autograd.Function
    Args:
        var: output Variable
        params: dict of (name, Variable) to add names to node that
            require grad (TODO: make optional)
    """
    if params is not None:
        assert isinstance(params.values()[0], Variable)
        param_map = {id(v): k for k, v in params.items()}

    node_attr = dict(style='filled',
                     shape='box',
                     align='left',
                     fontsize='12',
                     ranksep='0.1',
                     height='0.2')
    dot = Digraph(node_attr=node_attr, graph_attr=dict(size="12,12"))
    seen = set()

    def size_to_str(size):
        return '('+(', ').join(['%d' % v for v in size])+')'

    def add_nodes(var):
        if var not in seen:
            if torch.is_tensor(var):
                dot.node(str(id(var)), size_to_str(var.size()), fillcolor='orange')
            elif hasattr(var, 'variable'):
                u = var.variable
                name = param_map[id(u)] if params is not None else ''
                node_name = '%s\n %s' % (name, size_to_str(u.size()))
                dot.node(str(id(var)), node_name, fillcolor='lightblue')
            else:
                dot.node(str(id(var)), str(type(var).__name__))
            seen.add(var)
            if hasattr(var, 'next_functions'):
                for u in var.next_functions:
                    if u[0] is not None:
                        dot.edge(str(id(u[0])), str(id(var)))
                        add_nodes(u[0])
            if hasattr(var, 'saved_tensors'):
                for t in var.saved_tensors:
                    dot.edge(str(id(t)), str(id(var)))
                    add_nodes(t)
    add_nodes(var.grad_fn)
    return dot


def view_model(net, input_shape):
    x = Variable(torch.randn(1, *input_shape))
    y = net(x)
    g = make_dot(y)
    g.view()

    params = list(net.parameters())
    k = 0
    for i in params:
        l = 1
        print("layer parameters size:" + str(list(i.size())))
        for j in i.size():
            l *= j
        print("layer parameters:" + str(l))
        k = k + l
    print("total parameters:" + str(k))


if __name__ == '__main__':
    net = CNN()
    view_model(net)
    # x = Variable(torch.randn(1, 1, 28, 28))
    # y = net(x)
    # g = make_dot(y)
    # g.view()
    #
    # params = list(net.parameters())
    # k = 0
    # for i in params:
    #     l = 1
    #     print("layer parameters:" + str(list(i.size())))
    #     for j in i.size():
    #         l *= j
    #     print("layer parameters:" + str(l))
    #     k = k + l
    # print("total parameters:" + str(k))

2.6 utils/visualizer.py

下面这个脚本，主要通过visdom在线画图。其中有两个方法，display_current_results方法会根据输入的xy坐标画图，display_roc会根据y_true, y_pred计算roc曲线的横纵坐标fpr, tpr，然后画图。

import visdom
import time
import numpy as np
from matplotlib import pyplot as plt
from sklearn.metrics import roc_curve


class Visualizer(object):

    def __init__(self, env='default', **kwargs):
        self.vis = visdom.Visdom(env=env, **kwargs)
        self.vis.close()

        self.iters = {}
        self.lines = {}

    def display_current_results(self, iters, x, name='train_loss'):
        if name not in self.iters:
            self.iters[name] = []

        if name not in self.lines:
            self.lines[name] = []

        self.iters[name].append(iters)
        self.lines[name].append(x)

        self.vis.line(X=np.array(self.iters[name]),
                      Y=np.array(self.lines[name]),
                      win=name,
                      opts=dict(legend=[name], title=name))

    def display_roc(self, y_true, y_pred):
        fpr, tpr, ths = roc_curve(y_true, y_pred)
        self.vis.line(X=fpr,
                      Y=tpr,
                      # win='roc',
                      opts=dict(legend=['roc'],
                                title='roc'))

2.7 train.py

包含以下几个步骤：

（1）初始化数据读取dataloader（训练与测试集）

（2）初始化backbone模型

（3）初始化衡量的metric

（4）初始化优化器与学习率

（5）模型训练

（6）可视化训练过程

（7）模型测试

（8）可视化测试过程

from __future__ import print_function
import os
from data import Dataset
import torch
from torch.utils import data
import torch.nn.functional as F
from models import *
import torchvision
from utils import Visualizer, view_model
import torchVisualizer
import numpy as np
import random
import time
from config import Config
from torch.nn import DataParallel
from torch.optim.lr_scheduler import StepLR
from test import *


def save_model(model, save_path, name, iter_cnt)Visualizer:
    save_name = os.path.join(save_path, name + '_' + str(iter_cnt) + '.pth')
    torch.save(model.state_dict(), save_name)
    return save_name


if __name__ == '__main__':

    opt = Config()
    if opt.display:
        visualizer = Visualizer()
    device = torch.device("cuda")

    train_dataset = Dataset(opt.train_root, opt.train_list, phase='train', input_shape=opt.input_shape)
    trainloader = data.DataLoader(train_dataset,
                                  batch_size=opt.train_batch_size,
                                  shuffle=True,
                                  num_workers=opt.num_workers)

    identity_list = get_lfw_list(opt.lfw_test_list)
    img_paths = [os.path.join(opt.lfw_root, each) for each in identity_list]

    print('{} train iters per epoch:'.format(len(trainloader)))

    if opt.loss == 'focal_loss':
        criterion = FocalLoss(gamma=2)
    else:
        criterion = torch.nn.CrossEntropyLoss()

    if opt.backbone == 'resnet18':
        model = resnet_face18(use_se=opt.use_se)
    elif opt.backbone == 'resnet34':
        model = resnet34()
    elif opt.backbone == 'resnet50':
        model = resnet50()

    if opt.metric == 'add_margin':
        metric_fc = AddMarginProduct(512, opt.num_classes, s=30, m=0.35)
    elif opt.metric == 'arc_margin':
        metric_fc = ArcMarginProduct(512, opt.num_classes, s=30, m=0.5, easy_margin=opt.easy_margin)
    elif opt.metric == 'sphere':
        metric_fc = SphereProduct(512, opt.num_classes, m=4)
    else:
        metric_fc = nn.Linear(512, opt.num_classes)

    # view_model(model, opt.input_shape)
    print(model)
    model.to(device)
    model = DataParallel(model)
    metric_fc.to(device)
    metric_fc = DataParallel(metric_fc)

    if opt.optimizer == 'sgd':
        optimizer = torch.optim.SGD([{'params': model.parameters()}, {'params': metric_fc.parameters()}],
                                    lr=opt.lr, weight_decay=opt.weight_decay)
    else:
        optimizer = torch.optim.Adam([{'params': model.parameters()}, {'params': metric_fc.parameters()}],
                                     lr=opt.lr, weight_decay=opt.weight_decay)
    scheduler = StepLR(optimizer, step_size=opt.lr_step, gamma=0.1)

    start = time.time()
    for i in range(opt.max_epoch):
        scheduler.step()

        model.train()
        for ii, data in enumerate(trainloader):
            data_input, label = data
            data_input = data_input.to(device)
            label = label.to(device).long()
            feature = model(data_input)
            output = metric_fc(feature, label)
            loss = criterion(output, label)
            optimizer.zero_grad()
            loss.backward()
            optimizer.step()

            iters = i * len(trainloader) + ii

            if iters % opt.print_freq == 0:
                output = output.data.cpu().numpy()
                output = np.argmax(output, axis=1)
                label = label.data.cpu().numpy()
                # print(output)
                # print(label)
                acc = np.mean((output == label).astype(int))
                speed = opt.print_freq / (time.time() - start)
                time_str = time.asctime(time.localtime(time.time()))
                print('{} train epoch {} iter {} {} iters/s loss {} acc {}'.format(time_str, i, ii, speed, loss.item(), acc))
                if opt.display:
                    visualizer.display_current_results(iters, loss.item(), name='train_loss')
                    visualizer.display_current_results(iters, acc, name='train_acc')

                start = time.time()

        if i % opt.save_interval == 0 or i == opt.max_epoch:
            save_model(model, opt.checkpoints_path, opt.backbone, i)

        model.eval()
        acc = lfw_test(model, img_paths, identity_list, opt.lfw_test_list, opt.test_batch_size)
        if opt.display:
            visualizer.display_current_results(iters, acc, name='test_acc')