本案例参考课程：百度架构师手把手教深度学习的内容。 主要目的为练习Resnet动态图的PaddlePaddle实现。
本案例已经在AISTUDIO共享，链接为：
https://aistudio.baidu.com/aistudio/projectdetail/244766
数据集：
查看数据集图片 iChallenge-PM中既有病理性近视患者的眼底图片，也有非病理性近视患者的图片，命名规则如下：
病理性近视（PM）：文件名以P开头
非病理性近视（non-PM）：
高度近似（high myopia）：文件名以H开头
正常眼睛（normal）：文件名以N开头
我们将病理性患者的图片作为正样本，标签为1； 非病理性患者的图片作为负样本，标签为0。从数据集中选取两张图片，通过LeNet提取特征，构建分类器，对正负样本进行分类，并将图片显示出来。
ResNet
ResNet是2015年ImageNet比赛的冠军，将识别错误率降低到了3.6%，这个结果甚至超出了正常人眼识别的精度。
通过前面几个经典模型学习，我们可以发现随着深度学习的不断发展，模型的层数越来越多，网络结构也越来越复杂。那么是否加深网络结构，就一定会得到更好的效果呢？从理论上来说，假设新增加的层都是恒等映射，只要原有的层学出跟原模型一样的参数，那么深模型结构就能达到原模型结构的效果。换句话说，原模型的解只是新模型的解的子空间，在新模型解的空间里应该能找到比原模型解对应的子空间更好的结果。但是实践表明，增加网络的层数之后，训练误差往往不降反升。
Kaiming He等人提出了残差网络ResNet来解决上述问题，其基本思想如图6所示。
图6(a)：表示增加网络的时候，将x映射成y=F(x)y=F(x)y=F(x)输出。
图6(b)：对图6(a)作了改进，输出y=F(x)+xy=F(x) + xy=F(x)+x。这时不是直接学习输出特征y的表示，而是学习y−xy-xy−x。
如果想学习出原模型的表示，只需将F(x)的参数全部设置为0，则y=xy=xy=x是恒等映射。
F(x)=y−xF(x) = y - xF(x)=y−x也叫做残差项，如果x→yx\rightarrow yx→y的映射接近恒等映射，图6(b)中通过学习残差项也比图6(a)学习完整映射形式更加容易。
图6：残差块设计思想
图6(b)的结构是残差网络的基础，这种结构也叫做残差块（residual block）。输入x通过跨层连接，能更快的向前传播数据，或者向后传播梯度。残差块的具体设计方案如图7 所示，这种设计方案也成称作瓶颈结构（BottleNeck）。
图7：残差块结构示意图
下图表示出了ResNet-50的结构，一共包含49层卷积和1层全连接，所以被称为ResNet-50。
图8：ResNet-50模型网络结构示意图
ResNet-50的具体实现如下代码所示：
In[2]
import os
import numpy as np
import matplotlib.pyplot as plt
%matplotlib inline
from PIL import Image
DATADIR = '/home/aistudio/work/palm/PALM-Training400/PALM-Training400'
# 文件名以N开头的是正常眼底图片，以P开头的是病变眼底图片
file1 = 'N0012.jpg'
file2 = 'P0095.jpg'
# 读取图片
img1 = Image.open(os.path.join(DATADIR, file1))
img1 = np.array(img1)
img2 = Image.open(os.path.join(DATADIR, file2))
img2 = np.array(img2)
# 画出读取的图片
plt.figure(figsize=(16, 8))
f = plt.subplot(121)
f.set_title('Normal', fontsize=20)
plt.imshow(img1)
f = plt.subplot(122)
f.set_title('PM', fontsize=20)
plt.imshow(img2)
plt.show()
In[4]
# 查看图片形状
img1.shape, img2.shape
((2056, 2124, 3), (2056, 2124, 3))
In[3]
#定义数据读取器
import cv2
import random
import numpy as np
# 对读入的图像数据进行预处理
def transform_img(img):
# 将图片尺寸缩放道 224x224
img = cv2.resize(img, (224, 224))
# 读入的图像数据格式是[H, W, C]
# 使用转置操作将其变成[C, H, W]
img = np.transpose(img, (2,0,1))
img = img.astype('float32')
# 将数据范围调整到[-1.0, 1.0]之间
img = img / 255.
img = img * 2.0 - 1.0
return img
# 定义训练集数据读取器
def data_loader(datadir, batch_size=10, mode = 'train'):
# 将datadir目录下的文件列出来，每条文件都要读入
filenames = os.listdir(datadir)
def reader():
if mode == 'train':
# 训练时随机打乱数据顺序
random.shuffle(filenames)
batch_imgs = []
batch_labels = []
for name in filenames:
filepath = os.path.join(datadir, name)
img = cv2.imread(filepath)
img = transform_img(img)
if name[0] == 'H' or name[0] == 'N':
# H开头的文件名表示高度近似，N开头的文件名表示正常视力
# 高度近视和正常视力的样本，都不是病理性的，属于负样本，标签为0
label = 0
elif name[0] == 'P':
# P开头的是病理性近视，属于正样本，标签为1
label = 1
else:
raise('Not excepted file name')
# 每读取一个样本的数据，就将其放入数据列表中
batch_imgs.append(img)
batch_labels.append(label)
if len(batch_imgs) == batch_size:
# 当数据列表的长度等于batch_size的时候，
# 把这些数据当作一个mini-batch，并作为数据生成器的一个输出
imgs_array = np.array(batch_imgs).astype('float32')
labels_array = np.array(batch_labels).astype('float32').reshape(-1, 1)
yield imgs_array, labels_array
batch_imgs = []
batch_labels = []
if len(batch_imgs) > 0:
# 剩余样本数目不足一个batch_size的数据，一起打包成一个mini-batch
imgs_array = np.array(batch_imgs).astype('float32')
labels_array = np.array(batch_labels).astype('float32').reshape(-1, 1)
yield imgs_array, labels_array
return reader
# 定义验证集数据读取器
def valid_data_loader(datadir, csvfile, batch_size=10, mode='valid'):
# 训练集读取时通过文件名来确定样本标签，验证集则通过csvfile来读取每个图片对应的标签
# 请查看解压后的验证集标签数据，观察csvfile文件里面所包含的内容
# csvfile文件所包含的内容格式如下，每一行代表一个样本，
# 其中第一列是图片id，第二列是文件名，第三列是图片标签，
# 第四列和第五列是Fovea的坐标，与分类任务无关
# ID,imgName,Label,Fovea_X,Fovea_Y
# 1,V0001.jpg,0,1157.74,1019.87
# 2,V0002.jpg,1,1285.82,1080.47
# 打开包含验证集标签的csvfile，并读入其中的内容
filelists = open(csvfile).readlines()
def reader():
batch_imgs = []
batch_labels = []
for line in filelists[1:]:
line = line.strip().split(',')
name = line[1]
label = int(line[2])
# 根据图片文件名加载图片，并对图像数据作预处理
filepath = os.path.join(datadir, name)
img = cv2.imread(filepath)
img = transform_img(img)
# 每读取一个样本的数据，就将其放入数据列表中
batch_imgs.append(img)
batch_labels.append(label)
if len(batch_imgs) == batch_size:
# 当数据列表的长度等于batch_size的时候，
# 把这些数据当作一个mini-batch，并作为数据生成器的一个输出
imgs_array = np.array(batch_imgs).astype('float32')
labels_array = np.array(batch_labels).astype('float32').reshape(-1, 1)
yield imgs_array, labels_array
batch_imgs = []
batch_labels = []
if len(batch_imgs) > 0:
# 剩余样本数目不足一个batch_size的数据，一起打包成一个mini-batch
imgs_array = np.array(batch_imgs).astype('float32')
labels_array = np.array(batch_labels).astype('float32').reshape(-1, 1)
yield imgs_array, labels_array
return reader
In[5]
# 查看数据形状
DATADIR = '/home/aistudio/work/palm/PALM-Training400/PALM-Training400'
train_loader = data_loader(DATADIR,
batch_size=10, mode='train')
data_reader = train_loader()
data = next(data_reader)
data[0].shape, data[1].shape
((10, 3, 224, 224), (10, 1))
In[6]
!pip install xlrd
import pandas as pd
df=pd.read_excel('/home/aistudio/work/palm/PALM-Validation-GT/PM_Label_and_Fovea_Location.xlsx')
df.to_csv('/home/aistudio/work/palm/PALM-Validation-GT/labels.csv',index=False)
Looking in indexes: https://pypi.mirrors.ustc.edu.cn/simple/
Collecting xlrd
Downloading https://mirrors.tuna.tsinghua.edu.cn/pypi/web/packages/b0/16/63576a1a001752e34bf8ea62e367997530dc553b689356b9879339cf45a4/xlrd-1.2.0-py2.py3-none-any.whl (103kB)
|████████████████████████████████| 112kB 9.2MB/s eta 0:00:01
Installing collected packages: xlrd
Successfully installed xlrd-1.2.0
In[7]
#训练和评估代码
import os
import random
import paddle
import paddle.fluid as fluid
import numpy as np
DATADIR = '/home/aistudio/work/palm/PALM-Training400/PALM-Training400'
DATADIR2 = '/home/aistudio/work/palm/PALM-Validation400'
CSVFILE = '/home/aistudio/work/palm/PALM-Validation-GT/labels.csv'
# 定义训练过程
def train(model):
with fluid.dygraph.guard():
print('start training ... ')
model.train()
epoch_num = 5
# 定义优化器
opt = fluid.optimizer.Momentum(learning_rate=0.001, momentum=0.9)
# 定义数据读取器，训练数据读取器和验证数据读取器
train_loader = data_loader(DATADIR, batch_size=10, mode='train')
valid_loader = valid_data_loader(DATADIR2, CSVFILE)
for epoch in range(epoch_num):
for batch_id, data in enumerate(train_loader()):
x_data, y_data = data
img = fluid.dygraph.to_variable(x_data)
label = fluid.dygraph.to_variable(y_data)
# 运行模型前向计算，得到预测值
logits = model(img)
# 进行loss计算
loss = fluid.layers.sigmoid_cross_entropy_with_logits(logits, label)
avg_loss = fluid.layers.mean(loss)
if batch_id % 10 == 0:
print("epoch: {}, batch_id: {}, loss is: {}".format(epoch, batch_id, avg_loss.numpy()))
# 反向传播，更新权重，清除梯度
avg_loss.backward()
opt.minimize(avg_loss)
model.clear_gradients()
model.eval()
accuracies = []
losses = []
for batch_id, data in enumerate(valid_loader()):
x_data, y_data = data
img = fluid.dygraph.to_variable(x_data)
label = fluid.dygraph.to_variable(y_data)
# 运行模型前向计算，得到预测值
logits = model(img)
# 二分类，sigmoid计算后的结果以0.5为阈值分两个类别
# 计算sigmoid后的预测概率，进行loss计算
pred = fluid.layers.sigmoid(logits)
loss = fluid.layers.sigmoid_cross_entropy_with_logits(logits, label)
# 计算预测概率小于0.5的类别
pred2 = pred * (-1.0) + 1.0
# 得到两个类别的预测概率，并沿第一个维度级联
pred = fluid.layers.concat([pred2, pred], axis=1)
acc = fluid.layers.accuracy(pred, fluid.layers.cast(label, dtype='int64'))
accuracies.append(acc.numpy())
losses.append(loss.numpy())
print("[validation] accuracy/loss: {}/{}".format(np.mean(accuracies), np.mean(losses)))
model.train()
# save params of model
fluid.save_dygraph(model.state_dict(), 'mnist')
# save optimizer state
fluid.save_dygraph(opt.state_dict(), 'mnist')
# 定义评估过程
def evaluation(model, params_file_path):
with fluid.dygraph.guard():
print('start evaluation .......')
#加载模型参数
model_state_dict, _ = fluid.load_dygraph(params_file_path)
model.load_dict(model_state_dict)
model.eval()
eval_loader = load_data('eval')
acc_set = []
avg_loss_set = []
for batch_id, data in enumerate(eval_loader()):
x_data, y_data = data
img = fluid.dygraph.to_variable(x_data)
label = fluid.dygraph.to_variable(y_data)
# 计算预测和精度
prediction, acc = model(img, label)
# 计算损失函数值
loss = fluid.layers.cross_entropy(input=prediction, label=label)
avg_loss = fluid.layers.mean(loss)
acc_set.append(float(acc.numpy()))
avg_loss_set.append(float(avg_loss.numpy()))
# 求平均精度
acc_val_mean = np.array(acc_set).mean()
avg_loss_val_mean = np.array(avg_loss_set).mean()
print('loss={}, acc={}'.format(avg_loss_val_mean, acc_val_mean))
ResNet-50的具体实现如下代码所示：
In[8]
# -*- coding:utf-8 -*-
# ResNet模型代码
import numpy as np
import paddle
import paddle.fluid as fluid
from paddle.fluid.layer_helper import LayerHelper
from paddle.fluid.dygraph.nn import Conv2D, Pool2D, BatchNorm, FC
from paddle.fluid.dygraph.base import to_variable
# ResNet中使用了BatchNorm层，在卷积层的后面加上BatchNorm以提升数值稳定性
# 定义卷积批归一化块
class ConvBNLayer(fluid.dygraph.Layer):
def __init__(self,
name_scope,
num_channels,
num_filters,
filter_size,
stride=1,
groups=1,
act=None):
"""
name_scope, 模块的名字
num_channels, 卷积层的输入通道数
num_filters, 卷积层的输出通道数
stride, 卷积层的步幅
groups, 分组卷积的组数，默认groups=1不使用分组卷积
act, 激活函数类型，默认act=None不使用激活函数
"""
super(ConvBNLayer, self).__init__(name_scope)
# 创建卷积层
self._conv = Conv2D(
self.full_name(),
num_filters=num_filters,
filter_size=filter_size,
stride=stride,
padding=(filter_size - 1) // 2,
groups=groups,
act=None,
bias_attr=False)
# 创建BatchNorm层
self._batch_norm = BatchNorm(self.full_name(), num_filters, act=act)
def forward(self, inputs):
y = self._conv(inputs)
y = self._batch_norm(y)
return y
# 定义残差块
# 每个残差块会对输入图片做三次卷积，然后跟输入图片进行短接
# 如果残差块中第三次卷积输出特征图的形状与输入不一致，则对输入图片做1x1卷积，将其输出形状调整成一致
class BottleneckBlock(fluid.dygraph.Layer):
def __init__(self,
name_scope,
num_channels,
num_filters,
stride,
shortcut=True):
super(BottleneckBlock, self).__init__(name_scope)
# 创建第一个卷积层 1x1
self.conv0 = ConvBNLayer(
self.full_name(),
num_channels=num_channels,
num_filters=num_filters,
filter_size=1,
act='relu')
# 创建第二个卷积层 3x3
self.conv1 = ConvBNLayer(
self.full_name(),
num_channels=num_filters,
num_filters=num_filters,
filter_size=3,
stride=stride,
act='relu')
# 创建第三个卷积 1x1，但输出通道数乘以4
self.conv2 = ConvBNLayer(
self.full_name(),
num_channels=num_filters,
num_filters=num_filters * 4,
filter_size=1,
act=None)
# 如果conv2的输出跟此残差块的输入数据形状一致，则shortcut=True
# 否则shortcut = False，添加1个1x1的卷积作用在输入数据上，使其形状变成跟conv2一致
if not shortcut:
self.short = ConvBNLayer(
self.full_name(),
num_channels=num_channels,
num_filters=num_filters * 4,
filter_size=1,
stride=stride)
self.shortcut = shortcut
self._num_channels_out = num_filters * 4
def forward(self, inputs):
y = self.conv0(inputs)
conv1 = self.conv1(y)
conv2 = self.conv2(conv1)
# 如果shortcut=True，直接将inputs跟conv2的输出相加
# 否则需要对inputs进行一次卷积，将形状调整成跟conv2输出一致
if self.shortcut:
short = inputs
else:
short = self.short(inputs)
y = fluid.layers.elementwise_add(x=short, y=conv2)
layer_helper = LayerHelper(self.full_name(), act='relu')
return layer_helper.append_activation(y)
# 定义ResNet模型
class ResNet(fluid.dygraph.Layer):
def __init__(self, name_scope, layers=50, class_dim=1):
"""
name_scope，模块名称
layers, 网络层数，可以是50, 101或者152
class_dim，分类标签的类别数
"""
super(ResNet, self).__init__(name_scope)
self.layers = layers
supported_layers = [50, 101, 152]
assert layers in supported_layers, \
"supported layers are {} but input layer is {}".format(supported_layers, layers)
if layers == 50:
#ResNet50包含多个模块，其中第2到第5个模块分别包含3、4、6、3个残差块
depth = [3, 4, 6, 3]
elif layers == 101:
#ResNet101包含多个模块，其中第2到第5个模块分别包含3、4、23、3个残差块
depth = [3, 4, 23, 3]
elif layers == 152:
#ResNet50包含多个模块，其中第2到第5个模块分别包含3、8、36、3个残差块
depth = [3, 8, 36, 3]
# 残差块中使用到的卷积的输出通道数
num_filters = [64, 128, 256, 512]
# ResNet的第一个模块，包含1个7x7卷积，后面跟着1个最大池化层
self.conv = ConvBNLayer(
self.full_name(),
num_channels=3,
num_filters=64,
filter_size=7,
stride=2,
act='relu')
self.pool2d_max = Pool2D(
self.full_name(),
pool_size=3,
pool_stride=2,
pool_padding=1,
pool_type='max')
# ResNet的第二到第五个模块c2、c3、c4、c5
self.bottleneck_block_list = []
num_channels = 64
for block in range(len(depth)):
shortcut = False
for i in range(depth[block]):
bottleneck_block = self.add_sublayer(
'bb_%d_%d' % (block, i),
BottleneckBlock(
self.full_name(),
num_channels=num_channels,
num_filters=num_filters[block],
stride=2 if i == 0 and block != 0 else 1, # c3、c4、c5将会在第一个残差块使用stride=2；其余所有残差块stride=1
shortcut=shortcut))
num_channels = bottleneck_block._num_channels_out
self.bottleneck_block_list.append(bottleneck_block)
shortcut = True
# 在c5的输出特征图上使用全局池化
self.pool2d_avg = Pool2D(
self.full_name(), pool_size=7, pool_type='avg', global_pooling=True)
# stdv用来作为全连接层随机初始化参数的方差
import math
stdv = 1.0 / math.sqrt(2048 * 1.0)
# 创建全连接层，输出大小为类别数目
self.out = FC(self.full_name(),
size=class_dim,
param_attr=fluid.param_attr.ParamAttr(
initializer=fluid.initializer.Uniform(-stdv, stdv)))
def forward(self, inputs):
y = self.conv(inputs)
y = self.pool2d_max(y)
for bottleneck_block in self.bottleneck_block_list:
y = bottleneck_block(y)
y = self.pool2d_avg(y)
y = self.out(y)
return y
In[9]
with fluid.dygraph.guard():
model = ResNet("ResNet")
train(model)
start training ...
epoch: 0, batch_id: 0, loss is: [0.83079195]
epoch: 0, batch_id: 10, loss is: [0.5477183]
epoch: 0, batch_id: 20, loss is: [0.87052524]
epoch: 0, batch_id: 30, loss is: [1.0255078]
[validation] accuracy/loss: 0.7450000047683716/0.5235034823417664
epoch: 1, batch_id: 0, loss is: [0.41455013]
epoch: 1, batch_id: 10, loss is: [0.54812586]
epoch: 1, batch_id: 20, loss is: [0.17374663]
epoch: 1, batch_id: 30, loss is: [0.30293828]
[validation] accuracy/loss: 0.887499988079071/0.27671539783477783
epoch: 2, batch_id: 0, loss is: [0.38499922]
epoch: 2, batch_id: 10, loss is: [0.29150736]
epoch: 2, batch_id: 20, loss is: [0.3396409]
[validation] accuracy/loss: 0.9274999499320984/0.17061272263526917
epoch: 3, batch_id: 0, loss is: [0.06969612]
epoch: 3, batch_id: 10, loss is: [0.0861987]
epoch: 3, batch_id: 20, loss is: [0.05332329]
epoch: 3, batch_id: 30, loss is: [0.46470308]
[validation] accuracy/loss: 0.9375/0.20805077254772186
epoch: 4, batch_id: 0, loss is: [0.38617897]
epoch: 4, batch_id: 10, loss is: [0.16854036]
epoch: 4, batch_id: 20, loss is: [0.05454079]
epoch: 4, batch_id: 30, loss is: [0.32432565]
[validation] accuracy/loss: 0.8600000143051147/0.3488900661468506
来源：oschina
链接：https://my.oschina.net/u/4438602/blog/3157514