- 🍨 本文为🔗365天深度学习训练营 中的学习记录博客
- 🍖 原作者：K同学啊
前言
- 如果说最经典的神经网络，
ResNet肯定是一个，从ResNet发布后，很多人做了修改，denseNet网络无疑是最成功的一个，它采用密集型连接，将通道数连接在一起； - 本文是基于上一篇复现DenseNet121模型，做一个乳腺癌图像识别，效果还行，准确率0.9+;
- CNN经典网络之“DenseNet”简介，源码研究与复现(pytorch)： https://blog.csdn.net/weixin_74085818/article/details/146102290?spm=1001.2014.3001.5501
- 欢迎收藏 + 关注，本人将会持续更新
1、导入数据
1、导入库
import torch
import torch.nn as nn
import torchvision
import numpy as np
import os, PIL, pathlib
from collections import OrderedDict
import re
from torch.hub import load_state_dict_from_url
# 设置设备
device = "cuda" if torch.cuda.is_available() else "cpu"
device
'cuda'
2、查看数据信息和导入数据
data_dir = "./data/"
data_dir = pathlib.Path(data_dir)
# 类别数量
classnames = [str(path).split("\\")[0] for path in os.listdir(data_dir)]
classnames
['0', '1']
3、展示数据
import matplotlib.pylab as plt
from PIL import Image
# 获取文件名称
data_path_name = "./data/0/" # 不患病的
data_path_list = [f for f in os.listdir(data_path_name) if f.endswith(('jpg', 'png'))]
# 创建画板
fig, axes = plt.subplots(2, 8, figsize=(16, 6))
for ax, img_file in zip(axes.flat, data_path_list):
path_name = os.path.join(data_path_name, img_file)
img = Image.open(path_name) # 打开
# 显示
ax.imshow(img)
ax.axis('off')
plt.show()
4、数据导入
from torchvision import transforms, datasets
# 数据统一格式
img_height = 224
img_width = 224
data_tranforms = transforms.Compose([
transforms.Resize([img_height, img_width]),
transforms.ToTensor(),
transforms.Normalize( # 归一化
mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225]
)
])
# 加载所有数据
total_data = datasets.ImageFolder(root="./data/", transform=data_tranforms)
5、数据划分
# 大小 8 : 2
train_size = int(len(total_data) * 0.8)
test_size = len(total_data) - train_size
train_data, test_data = torch.utils.data.random_split(total_data, [train_size, test_size])
6、动态加载数据
batch_size = 64
train_dl = torch.utils.data.DataLoader(
train_data,
batch_size=batch_size,
shuffle=True
)
test_dl = torch.utils.data.DataLoader(
test_data,
batch_size=batch_size,
shuffle=False
)
# 查看数据维度
for data, labels in train_dl:
print("data shape[N, C, H, W]: ", data.shape)
print("labels: ", labels)
break
data shape[N, C, H, W]: torch.Size([64, 3, 224, 224])
labels: tensor([1, 1, 0, 0, 1, 1, 0, 0, 0, 1, 1, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1,
1, 0, 1, 1, 0, 0, 1, 1, 1, 0, 0, 0, 0, 1, 0, 1, 0, 0, 1, 0, 0, 0, 0, 0,
1, 0, 0, 1, 1, 1, 1, 1, 0, 0, 0, 1, 1, 0, 0, 1])
2、构建DenseNet121网络
import torch.nn.functional as F
# 实现DenseBlock中的部件：DenseLayer
'''
1、BN + ReLU: 处理部分，首先进行归一化，然后在用激活函数ReLU
2、Bottlenck Layer：称为瓶颈层，这个层在yolov5中常用，但是yolov5中主要用于特征提取+维度降维，这里采用1 * 1卷积核 + 3 * 3的卷积核进行卷积操作，目的：减少输入输入特征维度
3、BN + ReLU：对 瓶颈层 数据进行归一化，ReLU激活函数，归一化可以确保梯度下降的时候较为平稳
4、3 * 3 生成新的特征图
'''
class _DenseLayer(nn.Sequential):
def __init__(self, num_input_features, growth_rate, bn_size, drop_rate):
'''
num_input_features: 输入特征数，也就是通道数，在DenseNet中，每一层都会接受之前层的输出作为输入，故，这个数值通常会随着网络深度增加而增加
growth_rate: 增长率，这个是 DenseNet的核心概念，决定了每一层为全局状态贡献的特征数量，他的用处主要在于决定了中间瓶颈层的输出通道，需要结合代码去研究
bn_size: 瓶颈层中输出通道大小，含义：在使用1 * 1卷积核去提取特征数时，目标通道需要扩展到growth_rate的多少倍倍数， bn_size * growth_rate(输出维度)
drop_rate: 使用Dropout的参数
'''
super(_DenseLayer, self).__init__()
self.add_module("norm1", nn.BatchNorm2d(num_input_features))
self.add_module("relu1", nn.ReLU(inplace=True))
# 输出维度： bn_size * growth_rate, 1 * 1卷积核，步伐为1，只起到特征提取作用
self.add_module("conv1", nn.Conv2d(num_input_features, bn_size * growth_rate, stride=1, kernel_size=1, bias=False))
self.add_module("norm2", nn.BatchNorm2d(bn_size * growth_rate))
self.add_module("relu2", nn.ReLU(inplace=True))
# 输出通道：growth_rate, 维度计算：不变
self.add_module("conv2", nn.Conv2d(bn_size * growth_rate, growth_rate, stride=1, kernel_size=3, padding=1, bias=False))
self.drop_rate = drop_rate
def forward(self, x):
new_features = super(_DenseLayer, self).forward(x) # 传播
if self.drop_rate > 0:
new_features = F.dropout(new_features, p=self.drop_rate, training=self.training) # self.training 继承nn.Sequential，是否训练模式
# 模型融合，即，特征通道融合，形成新的特征图
return torch.cat([x, new_features], dim=1) # (N, C, H, W) # 即 C1 + C2，通道上融合
'''
DenseNet网络核心由DenseBlock模块组成，DenseBlock网络由DenseLayer组成，从 DenseLayer 可以看出，DenseBlock是
密集连接，每一层的输入不仅包含前一层的输出，还包含网络中所有之前层的输出
'''
# 构建DenseBlock模块, 通过上图
class _DenseBlock(nn.Sequential):
# num_layers 几层DenseLayer模块
def __init__(self, num_layers, num_input_features, bn_size, growth_rate, drop_rate):
super(_DenseBlock, self).__init__()
for i in range(num_layers):
layer = _DenseLayer(num_input_features + i * growth_rate, growth_rate, bn_size, drop_rate)
self.add_module("denselayer%d" % (i + 1), layer)
# Transition层，用于维度压缩
# 组成：一个卷积层 + 一个池化层
class _Transition(nn.Sequential):
def __init__(self, num_init_features, num_out_features):
super(_Transition, self).__init__()
self.add_module("norm", nn.BatchNorm2d(num_init_features))
self.add_module("relu", nn.ReLU(inplace=True))
self.add_module("conv", nn.Conv2d(num_init_features, num_out_features, kernel_size=1, stride=1, bias=False))
# 降维
self.add_module("pool", nn.AvgPool2d(2, stride=2))
# 搭建DenseNet网络
class DenseNet(nn.Module):
def __init__(self, growth_rate=32, block_config=(6, 12, 24, 16), num_init_features=64, bn_size=4,
compression_rate=0.5, drop_rate=0.5, num_classes=1000):
'''
growth_rate、num_init_features、num_init_features、drop_rate 和denselayer一样
block_config : 参数在 DenseNet 架构中用于指定每个 Dense Block 中包含的层数, 如：
DenseNet-121: block_config=(6, 12, 24, 16) 表示第一个 Dense Block 包含 6 层，第二个包含 12 层，第三个包含 24 层，第四个包含 16 层。
DenseNet-169: block_config=(6, 12, 32, 32)
DenseNet-201: block_config=(6, 12, 48, 32)
DenseNet-264: block_config=(6, 12, 64, 48)
compression_rate: 压缩维度, DenseNet 中用于 Transition Layer（过渡层）的一个重要参数，它控制了从一个 Dense Block 到下一个 Dense Block 之间特征维度的压缩程度
'''
super(DenseNet, self).__init__()
# 第一层卷积
# OrderedDict，让模型层有序排列
self.features = nn.Sequential(OrderedDict([
# 输出维度：((w - k + 2 * p) / s) + 1
("conv0", nn.Conv2d(3, num_init_features, kernel_size=7, stride=2, padding=3, bias=False)),
("norm0", nn.BatchNorm2d(num_init_features)),
("relu0", nn.ReLU(inplace=True)),
("pool0", nn.MaxPool2d(3, stride=2, padding=1)) # 降维
]))
# 搭建DenseBlock层
num_features = num_init_features
# num_layers: 层数
for i, num_layers in enumerate(block_config):
block = _DenseBlock(num_layers, num_features, bn_size, growth_rate, drop_rate)
# nn.Module 中features封装了nn.Sequential
self.features.add_module("denseblock%d" % (i + 1), block)
'''
# 这个计算反映了 DenseNet 中的一个关键特性：每一层输出的特征图（即新增加的通道数）由 growth_rate 决定，
# 并且这些新生成的特征图会被传递给该 Dense Block 中的所有后续层以及下一个 Dense Block。
'''
num_features += num_layers * growth_rate # 叠加，每一次叠加
# 判断是否需要使用Transition层
if i != len(block_config) - 1:
transition = _Transition(num_features, int(num_features*compression_rate)) # compression_rate 作用
self.features.add_module("transition%d" % (i + 1), transition)
num_features = int(num_features*compression_rate) # 更新维度
# 最后一层
self.features.add_module("norm5", nn.BatchNorm2d(num_features))
self.features.add_module("relu5", nn.ReLU(inplace=True))
# 分类层
self.classifier = nn.Linear(num_features, num_classes)
# params initialization
for m in self.modules():
if isinstance(m, nn.Conv2d):
'''
如果当前模块是一个二维卷积层 (nn.Conv2d)，那么它的权重 (m.weight) 将通过 Kaiming 正态分布 (kaiming_normal_) 进行初始化。
这种初始化方式特别适合与ReLU激活函数一起使用，有助于缓解深度网络中的梯度消失问题，促进有效的训练。
'''
nn.init.kaiming_normal_(m.weight)
elif isinstance(m, nn.BatchNorm2d):
'''
对于二维批归一化层 (nn.BatchNorm2d)，偏置项 (m.bias) 被初始化为0，而尺度因子 (m.weight) 被初始化为1。
这意味着在没有数据经过的情况下，批归一化层不会对输入进行额外的缩放或偏移，保持输入不变。
'''
nn.init.constant_(m.bias, 0)
nn.init.constant_(m.weight, 1)
elif isinstance(m, nn.Linear):
'''
对于全连接层 (nn.Linear)，只对其偏置项 (m.bias) 进行了初始化，设置为0'''
nn.init.constant_(m.bias, 0)
def forward(self, x):
features = self.features(x)
out = F.avg_pool2d(features, 7, stride=1).view(x.size(0), -1)
out = self.classifier(out)
return out
model = DenseNet(num_init_features=64, growth_rate=32, block_config=(6, 12, 12, 16))
model.to(device)
DenseNet(
(features): Sequential(
(conv0): Conv2d(3, 64, kernel_size=(7, 7), stride=(2, 2), padding=(3, 3), bias=False)
(norm0): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu0): ReLU(inplace=True)
(pool0): MaxPool2d(kernel_size=3, stride=2, padding=1, dilation=1, ceil_mode=False)
(denseblock1): _DenseBlock(
(denselayer1): _DenseLayer(
(norm1): BatchNorm2d(64, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu1): ReLU(inplace=True)
(conv1): Conv2d(64, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu2): ReLU(inplace=True)
(conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
)
(denselayer2): _DenseLayer(
(norm1): BatchNorm2d(96, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu1): ReLU(inplace=True)
(conv1): Conv2d(96, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu2): ReLU(inplace=True)
(conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
)
(denselayer3): _DenseLayer(
(norm1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu1): ReLU(inplace=True)
(conv1): Conv2d(128, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu2): ReLU(inplace=True)
(conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
)
(denselayer4): _DenseLayer(
(norm1): BatchNorm2d(160, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu1): ReLU(inplace=True)
(conv1): Conv2d(160, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu2): ReLU(inplace=True)
(conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
)
(denselayer5): _DenseLayer(
(norm1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu1): ReLU(inplace=True)
(conv1): Conv2d(192, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu2): ReLU(inplace=True)
(conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
)
(denselayer6): _DenseLayer(
(norm1): BatchNorm2d(224, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu1): ReLU(inplace=True)
(conv1): Conv2d(224, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu2): ReLU(inplace=True)
(conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
)
)
(transition1): _Transition(
(norm): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu): ReLU(inplace=True)
(conv): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(pool): AvgPool2d(kernel_size=2, stride=2, padding=0)
)
(denseblock2): _DenseBlock(
(denselayer1): _DenseLayer(
(norm1): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu1): ReLU(inplace=True)
(conv1): Conv2d(128, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu2): ReLU(inplace=True)
(conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
)
(denselayer2): _DenseLayer(
(norm1): BatchNorm2d(160, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu1): ReLU(inplace=True)
(conv1): Conv2d(160, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu2): ReLU(inplace=True)
(conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
)
(denselayer3): _DenseLayer(
(norm1): BatchNorm2d(192, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu1): ReLU(inplace=True)
(conv1): Conv2d(192, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu2): ReLU(inplace=True)
(conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
)
(denselayer4): _DenseLayer(
(norm1): BatchNorm2d(224, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu1): ReLU(inplace=True)
(conv1): Conv2d(224, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu2): ReLU(inplace=True)
(conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
)
(denselayer5): _DenseLayer(
(norm1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu1): ReLU(inplace=True)
(conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu2): ReLU(inplace=True)
(conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
)
(denselayer6): _DenseLayer(
(norm1): BatchNorm2d(288, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu1): ReLU(inplace=True)
(conv1): Conv2d(288, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu2): ReLU(inplace=True)
(conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
)
(denselayer7): _DenseLayer(
(norm1): BatchNorm2d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu1): ReLU(inplace=True)
(conv1): Conv2d(320, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu2): ReLU(inplace=True)
(conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
)
(denselayer8): _DenseLayer(
(norm1): BatchNorm2d(352, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu1): ReLU(inplace=True)
(conv1): Conv2d(352, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu2): ReLU(inplace=True)
(conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
)
(denselayer9): _DenseLayer(
(norm1): BatchNorm2d(384, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu1): ReLU(inplace=True)
(conv1): Conv2d(384, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu2): ReLU(inplace=True)
(conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
)
(denselayer10): _DenseLayer(
(norm1): BatchNorm2d(416, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu1): ReLU(inplace=True)
(conv1): Conv2d(416, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu2): ReLU(inplace=True)
(conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
)
(denselayer11): _DenseLayer(
(norm1): BatchNorm2d(448, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu1): ReLU(inplace=True)
(conv1): Conv2d(448, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu2): ReLU(inplace=True)
(conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
)
(denselayer12): _DenseLayer(
(norm1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu1): ReLU(inplace=True)
(conv1): Conv2d(480, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu2): ReLU(inplace=True)
(conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
)
)
(transition2): _Transition(
(norm): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu): ReLU(inplace=True)
(conv): Conv2d(512, 256, kernel_size=(1, 1), stride=(1, 1), bias=False)
(pool): AvgPool2d(kernel_size=2, stride=2, padding=0)
)
(denseblock3): _DenseBlock(
(denselayer1): _DenseLayer(
(norm1): BatchNorm2d(256, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu1): ReLU(inplace=True)
(conv1): Conv2d(256, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu2): ReLU(inplace=True)
(conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
)
(denselayer2): _DenseLayer(
(norm1): BatchNorm2d(288, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu1): ReLU(inplace=True)
(conv1): Conv2d(288, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu2): ReLU(inplace=True)
(conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
)
(denselayer3): _DenseLayer(
(norm1): BatchNorm2d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu1): ReLU(inplace=True)
(conv1): Conv2d(320, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu2): ReLU(inplace=True)
(conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
)
(denselayer4): _DenseLayer(
(norm1): BatchNorm2d(352, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu1): ReLU(inplace=True)
(conv1): Conv2d(352, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu2): ReLU(inplace=True)
(conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
)
(denselayer5): _DenseLayer(
(norm1): BatchNorm2d(384, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu1): ReLU(inplace=True)
(conv1): Conv2d(384, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu2): ReLU(inplace=True)
(conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
)
(denselayer6): _DenseLayer(
(norm1): BatchNorm2d(416, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu1): ReLU(inplace=True)
(conv1): Conv2d(416, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu2): ReLU(inplace=True)
(conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
)
(denselayer7): _DenseLayer(
(norm1): BatchNorm2d(448, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu1): ReLU(inplace=True)
(conv1): Conv2d(448, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu2): ReLU(inplace=True)
(conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
)
(denselayer8): _DenseLayer(
(norm1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu1): ReLU(inplace=True)
(conv1): Conv2d(480, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu2): ReLU(inplace=True)
(conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
)
(denselayer9): _DenseLayer(
(norm1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu1): ReLU(inplace=True)
(conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu2): ReLU(inplace=True)
(conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
)
(denselayer10): _DenseLayer(
(norm1): BatchNorm2d(544, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu1): ReLU(inplace=True)
(conv1): Conv2d(544, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu2): ReLU(inplace=True)
(conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
)
(denselayer11): _DenseLayer(
(norm1): BatchNorm2d(576, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu1): ReLU(inplace=True)
(conv1): Conv2d(576, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu2): ReLU(inplace=True)
(conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
)
(denselayer12): _DenseLayer(
(norm1): BatchNorm2d(608, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu1): ReLU(inplace=True)
(conv1): Conv2d(608, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu2): ReLU(inplace=True)
(conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
)
)
(transition3): _Transition(
(norm): BatchNorm2d(640, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu): ReLU(inplace=True)
(conv): Conv2d(640, 320, kernel_size=(1, 1), stride=(1, 1), bias=False)
(pool): AvgPool2d(kernel_size=2, stride=2, padding=0)
)
(denseblock4): _DenseBlock(
(denselayer1): _DenseLayer(
(norm1): BatchNorm2d(320, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu1): ReLU(inplace=True)
(conv1): Conv2d(320, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu2): ReLU(inplace=True)
(conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
)
(denselayer2): _DenseLayer(
(norm1): BatchNorm2d(352, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu1): ReLU(inplace=True)
(conv1): Conv2d(352, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu2): ReLU(inplace=True)
(conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
)
(denselayer3): _DenseLayer(
(norm1): BatchNorm2d(384, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu1): ReLU(inplace=True)
(conv1): Conv2d(384, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu2): ReLU(inplace=True)
(conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
)
(denselayer4): _DenseLayer(
(norm1): BatchNorm2d(416, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu1): ReLU(inplace=True)
(conv1): Conv2d(416, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu2): ReLU(inplace=True)
(conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
)
(denselayer5): _DenseLayer(
(norm1): BatchNorm2d(448, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu1): ReLU(inplace=True)
(conv1): Conv2d(448, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu2): ReLU(inplace=True)
(conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
)
(denselayer6): _DenseLayer(
(norm1): BatchNorm2d(480, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu1): ReLU(inplace=True)
(conv1): Conv2d(480, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu2): ReLU(inplace=True)
(conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
)
(denselayer7): _DenseLayer(
(norm1): BatchNorm2d(512, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu1): ReLU(inplace=True)
(conv1): Conv2d(512, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu2): ReLU(inplace=True)
(conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
)
(denselayer8): _DenseLayer(
(norm1): BatchNorm2d(544, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu1): ReLU(inplace=True)
(conv1): Conv2d(544, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu2): ReLU(inplace=True)
(conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
)
(denselayer9): _DenseLayer(
(norm1): BatchNorm2d(576, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu1): ReLU(inplace=True)
(conv1): Conv2d(576, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu2): ReLU(inplace=True)
(conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
)
(denselayer10): _DenseLayer(
(norm1): BatchNorm2d(608, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu1): ReLU(inplace=True)
(conv1): Conv2d(608, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu2): ReLU(inplace=True)
(conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
)
(denselayer11): _DenseLayer(
(norm1): BatchNorm2d(640, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu1): ReLU(inplace=True)
(conv1): Conv2d(640, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu2): ReLU(inplace=True)
(conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
)
(denselayer12): _DenseLayer(
(norm1): BatchNorm2d(672, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu1): ReLU(inplace=True)
(conv1): Conv2d(672, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu2): ReLU(inplace=True)
(conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
)
(denselayer13): _DenseLayer(
(norm1): BatchNorm2d(704, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu1): ReLU(inplace=True)
(conv1): Conv2d(704, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu2): ReLU(inplace=True)
(conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
)
(denselayer14): _DenseLayer(
(norm1): BatchNorm2d(736, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu1): ReLU(inplace=True)
(conv1): Conv2d(736, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu2): ReLU(inplace=True)
(conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
)
(denselayer15): _DenseLayer(
(norm1): BatchNorm2d(768, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu1): ReLU(inplace=True)
(conv1): Conv2d(768, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu2): ReLU(inplace=True)
(conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
)
(denselayer16): _DenseLayer(
(norm1): BatchNorm2d(800, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu1): ReLU(inplace=True)
(conv1): Conv2d(800, 128, kernel_size=(1, 1), stride=(1, 1), bias=False)
(norm2): BatchNorm2d(128, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu2): ReLU(inplace=True)
(conv2): Conv2d(128, 32, kernel_size=(3, 3), stride=(1, 1), padding=(1, 1), bias=False)
)
)
(norm5): BatchNorm2d(832, eps=1e-05, momentum=0.1, affine=True, track_running_stats=True)
(relu5): ReLU(inplace=True)
)
(classifier): Linear(in_features=832, out_features=1000, bias=True)
)
3、模型训练
1、构建训练集
def train(dataloader, model, loss_fn, optimizer):
size = len(dataloader.dataset)
batch_size = len(dataloader)
train_acc, train_loss = 0, 0
for X, y in dataloader:
X, y = X.to(device), y.to(device)
# 训练
pred = model(X)
loss = loss_fn(pred, y)
# 梯度下降法
optimizer.zero_grad()
loss.backward()
optimizer.step()
# 记录
train_loss += loss.item()
train_acc += (pred.argmax(1) == y).type(torch.float).sum().item()
train_acc /= size
train_loss /= batch_size
return train_acc, train_loss
2、构建测试集
def test(dataloader, model, loss_fn):
size = len(dataloader.dataset)
batch_size = len(dataloader)
test_acc, test_loss = 0, 0
with torch.no_grad():
for X, y in dataloader:
X, y = X.to(device), y.to(device)
pred = model(X)
loss = loss_fn(pred, y)
test_loss += loss.item()
test_acc += (pred.argmax(1) == y).type(torch.float).sum().item()
test_acc /= size
test_loss /= batch_size
return test_acc, test_loss
3、设置超参数
loss_fn = nn.CrossEntropyLoss() # 损失函数
learn_lr = 1e-4 # 超参数
optimizer = torch.optim.Adam(model.parameters(), lr=learn_lr) # 优化器
4、模型训练
通过实验发现，还是设置20轮次附件最好
import copy
train_acc = []
train_loss = []
test_acc = []
test_loss = []
epoches = 20
best_acc = 0 # 设置一个最佳准确率，作为最佳模型的判别指标
for i in range(epoches):
model.train()
epoch_train_acc, epoch_train_loss = train(train_dl, model, loss_fn, optimizer)
model.eval()
epoch_test_acc, epoch_test_loss = test(test_dl, model, loss_fn)
# 保存最佳模型到 best_model
if epoch_test_acc > best_acc:
best_acc = epoch_test_acc
best_model = copy.deepcopy(model)
train_acc.append(epoch_train_acc)
train_loss.append(epoch_train_loss)
test_acc.append(epoch_test_acc)
test_loss.append(epoch_test_loss)
# 获取当前的学习率
lr = optimizer.state_dict()['param_groups'][0]['lr']
# 输出
template = ('Epoch:{:2d}, Train_acc:{:.1f}%, Train_loss:{:.3f}, Test_acc:{:.1f}%, Test_loss:{:.3f}')
print(template.format(i + 1, epoch_train_acc*100, epoch_train_loss, epoch_test_acc*100, epoch_test_loss))
PATH = './best_model.pth' # 保存的参数文件名
torch.save(best_model.state_dict(), PATH)
print("Done")
Epoch: 1, Train_acc:79.3%, Train_loss:1.948, Test_acc:84.6%, Test_loss:1.079
Epoch: 2, Train_acc:85.3%, Train_loss:0.395, Test_acc:85.2%, Test_loss:0.721
Epoch: 3, Train_acc:87.3%, Train_loss:0.318, Test_acc:86.5%, Test_loss:0.526
Epoch: 4, Train_acc:89.0%, Train_loss:0.277, Test_acc:86.6%, Test_loss:0.494
Epoch: 5, Train_acc:89.0%, Train_loss:0.266, Test_acc:87.9%, Test_loss:0.400
Epoch: 6, Train_acc:89.6%, Train_loss:0.252, Test_acc:84.6%, Test_loss:0.524
Epoch: 7, Train_acc:90.3%, Train_loss:0.239, Test_acc:85.5%, Test_loss:0.445
Epoch: 8, Train_acc:90.2%, Train_loss:0.235, Test_acc:87.6%, Test_loss:0.359
Epoch: 9, Train_acc:90.0%, Train_loss:0.235, Test_acc:89.3%, Test_loss:0.298
Epoch:10, Train_acc:91.0%, Train_loss:0.220, Test_acc:89.5%, Test_loss:0.307
Epoch:11, Train_acc:90.8%, Train_loss:0.222, Test_acc:88.3%, Test_loss:0.316
Epoch:12, Train_acc:91.4%, Train_loss:0.210, Test_acc:83.3%, Test_loss:0.516
Epoch:13, Train_acc:91.5%, Train_loss:0.208, Test_acc:91.3%, Test_loss:0.247
Epoch:14, Train_acc:91.5%, Train_loss:0.206, Test_acc:90.1%, Test_loss:0.269
Epoch:15, Train_acc:92.0%, Train_loss:0.199, Test_acc:91.1%, Test_loss:0.242
Epoch:16, Train_acc:92.1%, Train_loss:0.194, Test_acc:89.4%, Test_loss:0.285
Epoch:17, Train_acc:92.4%, Train_loss:0.193, Test_acc:91.0%, Test_loss:0.229
Epoch:18, Train_acc:92.4%, Train_loss:0.188, Test_acc:88.0%, Test_loss:0.317
Epoch:19, Train_acc:92.7%, Train_loss:0.182, Test_acc:89.2%, Test_loss:0.285
Epoch:20, Train_acc:92.6%, Train_loss:0.182, Test_acc:78.5%, Test_loss:0.728
Done
5、结果可视化
import matplotlib.pyplot as plt
#隐藏警告
import warnings
warnings.filterwarnings("ignore") #忽略警告信息
epochs_range = range(epoches)
plt.figure(figsize=(12, 3))
plt.subplot(1, 2, 1)
plt.plot(epochs_range, train_acc, label='Training Accuracy')
plt.plot(epochs_range, test_acc, label='Test Accuracy')
plt.legend(loc='lower right')
plt.title('Training Accuracy')
plt.subplot(1, 2, 2)
plt.plot(epochs_range, train_loss, label='Training Loss')
plt.plot(epochs_range, test_loss, label='Test Loss')
plt.legend(loc='upper right')
plt.title('Training= Loss')
plt.show()
在20轮测试集准确率变化比较大，从跑的几次实验来看，这次是偶然事件，测试集损失率后面一直稳定在0.3附件，测试准确率一直在0.8、0.89、0.90附件徘徊
6、模型评估
# 将参数加载到model当中
best_model.load_state_dict(torch.load(PATH, map_location=device))
epoch_test_acc, epoch_test_loss = test(test_dl, best_model, loss_fn)
print(epoch_test_acc, epoch_test_loss)
0.9134651249533756 0.24670581874393283