第61步 深度学习图像识别：多分类建模（TensorFlow）

Jet45052023-08-27 10:23

基于WIN10的64位系统演示

一、写在前面

截至上期，我们一直都在做二分类的任务，无论是之前的机器学习任务，还是最近更新的图像分类任务。然而，在实际工作中，我们大概率需要进行多分类任务。例如肺部胸片可不仅仅能诊断肺结核，还有COVID-19、细菌性（病毒性）肺炎等等，这就涉及到图像识别的多分类任务。

本期以健康组、肺结核组、COVID-19组、细菌性（病毒性）肺炎组为数据集，构建Mobilenet多分类模型，原因还是因为它建模速度快。

同样，基于GPT-4辅助编程，改写过程见后面。

二、误判病例分析实战

使用胸片的数据集：肺结核病人和健康人的胸片的识别。其中，健康人900张，肺结核病人700张，COVID-19病人549张、细菌性（病毒性）肺炎组900张，分别存入单独的文件夹中。

（a）直接分享代码

python 复制代码 
######################################导入包###################################
from tensorflow import keras
import tensorflow as tf
from tensorflow.python.keras.layers import Dense, Flatten, Conv2D, MaxPool2D, Dropout, Activation, Reshape, Softmax, GlobalAveragePooling2D, BatchNormalization
from tensorflow.python.keras.layers.convolutional import Convolution2D, MaxPooling2D
from tensorflow.python.keras import Sequential
from tensorflow.python.keras import Model
from tensorflow.python.keras.optimizers import adam_v2
import numpy as np
import matplotlib.pyplot as plt
from tensorflow.python.keras.preprocessing.image import ImageDataGenerator, image_dataset_from_directory
from tensorflow.python.keras.layers.preprocessing.image_preprocessing import RandomFlip, RandomRotation, RandomContrast, RandomZoom, RandomTranslation
import os,PIL,pathlib
import warnings
#设置GPU
gpus = tf.config.list_physical_devices("GPU")

if gpus:
    gpu0 = gpus[0] #如果有多个GPU，仅使用第0个GPU
    tf.config.experimental.set_memory_growth(gpu0, True) #设置GPU显存用量按需使用
    tf.config.set_visible_devices([gpu0],"GPU")
    
warnings.filterwarnings("ignore")             #忽略警告信息
plt.rcParams['font.sans-serif'] = ['SimHei']  # 用来正常显示中文标签
plt.rcParams['axes.unicode_minus'] = False    # 用来正常显示负号


################################导入数据集#####################################
#1.导入数据
#1.导入数据
data_dir = "./MTB-1" # 修改了路径
data_dir = pathlib.Path(data_dir)
image_count = len(list(data_dir.glob('*/*')))
print("图片总数为：",image_count)

batch_size = 32
img_height = 100
img_width  = 100

train_ds = image_dataset_from_directory(
    data_dir,
    validation_split=0.2,
    subset="training",
    seed=12,
    image_size=(img_height, img_width),
    batch_size=batch_size)

val_ds = image_dataset_from_directory(
    data_dir,
    validation_split=0.2,
    subset="validation",
    seed=12,
    image_size=(img_height, img_width),
    batch_size=batch_size)

class_names = train_ds.class_names
print(class_names)
print(train_ds)


#2.检查数据
for image_batch, labels_batch in train_ds:
    print(image_batch.shape)
    print(labels_batch.shape)
    break

#3.配置数据
AUTOTUNE = tf.data.AUTOTUNE

def train_preprocessing(image,label):
    return (image/255.0,label)

train_ds = (
    train_ds.cache()
    .shuffle(800)
    .map(train_preprocessing)    # 这里可以设置预处理函数
#     .batch(batch_size)           # 在image_dataset_from_directory处已经设置了batch_size
    .prefetch(buffer_size=AUTOTUNE)
)

val_ds = (
    val_ds.cache()
    .map(train_preprocessing)    # 这里可以设置预处理函数
#     .batch(batch_size)         # 在image_dataset_from_directory处已经设置了batch_size
    .prefetch(buffer_size=AUTOTUNE)
)

#4. 数据可视化
plt.figure(figsize=(10, 8))  # 图形的宽为10高为5
plt.suptitle("数据展示")

class_names = ["COVID-19", "Normal", "Pneumonia", "Tuberculosis"] # 修改类别标签

for images, labels in train_ds.take(1):
    for i in range(15):
        plt.subplot(4, 5, i + 1)
        plt.xticks([])
        plt.yticks([])
        plt.grid(False)

        # 显示图片
        plt.imshow(images[i])
        # 显示标签
        plt.xlabel(class_names[labels[i]-1])

plt.show()

######################################数据增强函数################################

data_augmentation = Sequential([
  RandomFlip("horizontal_and_vertical"),
  RandomRotation(0.2),
  RandomContrast(1.0),
  RandomZoom(0.5,0.2),
  RandomTranslation(0.3,0.5),
])

def prepare(ds):
    ds = ds.map(lambda x, y: (data_augmentation(x, training=True), y), num_parallel_calls=AUTOTUNE)
    return ds
train_ds = prepare(train_ds)


###############################导入mobilenet_v2################################
#获取预训练模型对输入的预处理方法
from tensorflow.python.keras.applications import mobilenet_v2
from tensorflow.python.keras import Input, regularizers
IMG_SIZE = (img_height, img_width, 3)

base_model = mobilenet_v2.MobileNetV2(input_shape=IMG_SIZE, 
                                      include_top=False, #是否包含顶层的全连接层
                                      weights='imagenet')

inputs = Input(shape=IMG_SIZE)
#模型
x = base_model(inputs, training=False) #参数不变化
#全局池化
x = GlobalAveragePooling2D()(x)
#BatchNormalization
x = BatchNormalization()(x)
#Dropout
x = Dropout(0.8)(x)
#Dense
x = Dense(128, kernel_regularizer=regularizers.l2(0.1))(x)  # 全连接层减少到128，添加 L2 正则化
#BatchNormalization
x = BatchNormalization()(x)
#激活函数
x = Activation('relu')(x)
#输出层
outputs = Dense(4, kernel_regularizer=regularizers.l2(0.1))(x)  # 输出层神经元数量修改为4
#BatchNormalization
outputs = BatchNormalization()(outputs)
#激活函数
outputs = Activation('softmax')(outputs) # 激活函数修改为'softmax'
#整体封装
model = Model(inputs, outputs)
#打印模型结构
print(model.summary())
#############################编译模型#########################################
#定义优化器
from tensorflow.python.keras.optimizers import adam_v2, rmsprop_v2
#from tensorflow.python.keras.optimizer_v2.gradient_descent import SGD
optimizer = adam_v2.Adam()
#optimizer = SGD(learning_rate=0.001)
#optimizer = rmsprop_v2.RMSprop()

#常用的优化器
#all_classes = {
#      'adadelta': adadelta_v2.Adadelta,
#     'adagrad': adagrad_v2.Adagrad,
#     'adam': adam_v2.Adam,
#      'adamax': adamax_v2.Adamax,
#      'experimentaladadelta': adadelta_experimental.Adadelta,
#      'experimentaladagrad': adagrad_experimental.Adagrad,
#      'experimentaladam': adam_experimental.Adam,
#      'experimentalsgd': sgd_experimental.SGD,
#      'nadam': nadam_v2.Nadam,
#      'rmsprop': rmsprop_v2.RMSprop,

#编译模型
model.compile(optimizer=optimizer,
                loss='sparse_categorical_crossentropy', # 多分类问题
                metrics=['accuracy'])

#训练模型
from tensorflow.python.keras.callbacks import ModelCheckpoint, Callback, EarlyStopping, ReduceLROnPlateau, LearningRateScheduler

NO_EPOCHS = 50
PATIENCE  = 10
VERBOSE   = 1

# 设置动态学习率
annealer = LearningRateScheduler(lambda x: 1e-5 * 0.99 ** (x+NO_EPOCHS))

# 设置早停
earlystopper = EarlyStopping(monitor='loss', patience=PATIENCE, verbose=VERBOSE)

# 
checkpointer = ModelCheckpoint('mtb_4_jet_best_model_mobilenetv3samll.h5',
                                monitor='val_accuracy',
                                verbose=VERBOSE,
                                save_best_only=True,
                                save_weights_only=True)

train_model  = model.fit(train_ds,
                  epochs=NO_EPOCHS,
                  verbose=1,
                  validation_data=val_ds,
                  callbacks=[earlystopper, checkpointer, annealer])

#保存模型
model.save('mtb_4_jet_best_model_mobilenet.h5')
print("The trained model has been saved.")

from tensorflow.python.keras.models import load_model
train_model=load_model('mtb_4_jet_best_model_mobilenet.h5')
###########################Accuracy和Loss可视化#################################
import matplotlib.pyplot as plt

loss = train_model.history['loss']
acc = train_model.history['accuracy']
val_loss = train_model.history['val_loss']
val_acc = train_model.history['val_accuracy']
epoch = range(1, len(loss)+1)

fig, ax = plt.subplots(1, 2, figsize=(10,4))
ax[0].plot(epoch, loss, label='Train loss')
ax[0].plot(epoch, val_loss, label='Validation loss')
ax[0].set_xlabel('Epochs')
ax[0].set_ylabel('Loss')
ax[0].legend()
ax[1].plot(epoch, acc, label='Train acc')
ax[1].plot(epoch, val_acc, label='Validation acc')
ax[1].set_xlabel('Epochs')
ax[1].set_ylabel('Accuracy')
ax[1].legend()
#plt.show()
plt.savefig("loss-acc.pdf", dpi=300,format="pdf")

####################################混淆矩阵可视化#############################
import numpy as np
import matplotlib.pyplot as plt
from tensorflow.python.keras.models import load_model
from matplotlib.pyplot import imshow
from sklearn.metrics import classification_report, confusion_matrix
import seaborn as sns
import pandas as pd
import math
from sklearn.metrics import precision_recall_fscore_support, accuracy_score

# 定义一个绘制混淆矩阵图的函数
def plot_cm(labels, predictions, class_names):
    # 生成混淆矩阵
    conf_numpy = confusion_matrix(labels, predictions)
    # 将矩阵转化为 DataFrame
    conf_df = pd.DataFrame(conf_numpy, index=class_names ,columns=class_names)  
    
    plt.figure(figsize=(8,7))
    
    sns.heatmap(conf_df, annot=True, fmt="d", cmap="BuPu")
    
    plt.title('Confusion matrix',fontsize=15)
    plt.ylabel('Actual value',fontsize=14)
    plt.xlabel('Predictive value',fontsize=14)

val_pre   = []
val_label = []
for images, labels in val_ds:
    for image, label in zip(images, labels):
        img_array = tf.expand_dims(image, 0)
        prediction = model.predict(img_array)
        val_pre.append(np.argmax(prediction, axis=-1))
        val_label.append(label.numpy())  # 需要将标签转换为 numpy 数组

class_names = ['COVID-19', 'Normal', 'Pneumonia', 'Tuberculosis']  # 修改为你的类别名称
plot_cm(val_label, val_pre, class_names)
plt.savefig("val-cm.pdf", dpi=300,format="pdf")

precision_val, recall_val, f1_val, _ = precision_recall_fscore_support(val_label, val_pre, average='micro')
acc_val = accuracy_score(val_label, val_pre)
error_rate_val = 1 - acc_val

print("验证集的灵敏度（召回率）为：",recall_val, 
      "验证集的特异度为：",precision_val,  # 在多分类问题中，特异度定义不明确，这里我们使用精确度来代替
      "验证集的准确率为：",acc_val, 
      "验证集的错误率为：",error_rate_val,
      "验证集的F1为：",f1_val)

train_pre   = []
train_label = []
for images, labels in train_ds:
    for image, label in zip(images, labels):
        img_array = tf.expand_dims(image, 0)
        prediction = model.predict(img_array)
        train_pre.append(np.argmax(prediction, axis=-1))
        train_label.append(label.numpy())

plot_cm(train_label, train_pre, class_names)
plt.savefig("train-cm.pdf", dpi=300,format="pdf")

precision_train, recall_train, f1_train, _ = precision_recall_fscore_support(train_label, train_pre, average='micro')
acc_train = accuracy_score(train_label, train_pre)
error_rate_train = 1 - acc_train

print("训练集的灵敏度（召回率）为：",recall_train, 
      "训练集的特异度为：",precision_train,  # 在多分类问题中，特异度定义不明确，这里我们使用精确度来代替
      "训练集的准确率为：",acc_train, 
      "训练集的错误率为：",error_rate_train,
      "训练集的F1为：",f1_train)


################################模型性能参数计算################################
from sklearn import metrics

def test_accuracy_report(model):
    print(metrics.classification_report(val_label, val_pre, target_names=class_names)) 
    score = model.evaluate(val_ds, verbose=0)
    print('Loss function: %s, accuracy:' % score[0], score[1])
    
test_accuracy_report(model)

def train_accuracy_report(model):
    print(metrics.classification_report(train_label, train_pre, target_names=class_names)) 
    score = model.evaluate(train_ds, verbose=0)
    print('Loss function: %s, accuracy:' % score[0], score[1])
    
train_accuracy_report(model)

################################AUC曲线绘制####################################
from sklearn import metrics
from sklearn.preprocessing import LabelBinarizer
import numpy as np
import matplotlib.pyplot as plt
from tensorflow.python.keras.models import load_model
from matplotlib.pyplot import imshow
from sklearn.metrics import classification_report, confusion_matrix
import seaborn as sns
import pandas as pd
import math

def plot_roc(name, labels, predictions, **kwargs):
    fp, tp, _ = metrics.roc_curve(labels, predictions)

    plt.plot(fp, tp, label=name, linewidth=2, **kwargs)
    plt.xlabel('False positives rate')
    plt.ylabel('True positives rate')
    ax = plt.gca()
    ax.set_aspect('equal')

# 需要将标签进行one-hot编码
lb = LabelBinarizer()
lb.fit([0, 1, 2, 3])  # 训练标签编码器，这里设定有四个类别
n_classes = 4  # 类别数量

val_pre_auc   = []
val_label_auc = []

for images, labels in val_ds:
    for image, label in zip(images, labels):      
        img_array = tf.expand_dims(image, 0) 
        prediction_auc = model.predict(img_array)
        val_pre_auc.append(prediction_auc[0])
        val_label_auc.append(lb.transform([label])[0])  # 这里需要使用标签编码器进行编码

val_pre_auc = np.array(val_pre_auc)
val_label_auc = np.array(val_label_auc)

auc_score_val = [metrics.roc_auc_score(val_label_auc[:, i], val_pre_auc[:, i]) for i in range(n_classes)]


train_pre_auc   = []
train_label_auc = []

for images, labels in train_ds:
    for image, label in zip(images, labels):
        img_array_train = tf.expand_dims(image, 0) 
        prediction_auc = model.predict(img_array_train)
        train_pre_auc.append(prediction_auc[0])
        train_label_auc.append(lb.transform([label])[0])

train_pre_auc = np.array(train_pre_auc)
train_label_auc = np.array(train_label_auc)

auc_score_train = [metrics.roc_auc_score(train_label_auc[:, i], train_pre_auc[:, i]) for i in range(n_classes)]

for i in range(n_classes):
    plot_roc('validation AUC for class {0}: {1:.4f}'.format(i, auc_score_val[i]), val_label_auc[:, i] , val_pre_auc[:, i], color="red", linestyle='--')
    plot_roc('training AUC for class {0}: {1:.4f}'.format(i, auc_score_train[i]), train_label_auc[:, i], train_pre_auc[:, i], color="blue", linestyle='--')

plt.legend(loc='lower right')
plt.savefig("roc.pdf", dpi=300,format="pdf")

for i in range(n_classes):
    print("Class {0} 训练集的AUC值为：".format(i), auc_score_train[i], "验证集的AUC值为：", auc_score_val[i])


################################AUC曲线绘制-分开展示####################################
from sklearn import metrics
from sklearn.preprocessing import LabelBinarizer
import numpy as np
import matplotlib.pyplot as plt
from tensorflow.python.keras.models import load_model
from matplotlib.pyplot import imshow
from sklearn.metrics import classification_report, confusion_matrix
import seaborn as sns
import pandas as pd
import math

def plot_roc(ax, name, labels, predictions, **kwargs):
    fp, tp, _ = metrics.roc_curve(labels, predictions)
    ax.plot(fp, tp, label=name, linewidth=2, **kwargs)
    ax.plot([0, 1], [0, 1], color='orange', linestyle='--')
    ax.set_xlabel('False positives rate')
    ax.set_ylabel('True positives rate')
    ax.set_aspect('equal')

# 需要将标签进行one-hot编码
lb = LabelBinarizer()
lb.fit([0, 1, 2, 3])  # 训练标签编码器，这里设定有四个类别
n_classes = 4  # 类别数量

val_pre_auc   = []
val_label_auc = []

for images, labels in val_ds:
    for image, label in zip(images, labels):      
        img_array = tf.expand_dims(image, 0) 
        prediction_auc = model.predict(img_array)
        val_pre_auc.append(prediction_auc[0])
        val_label_auc.append(lb.transform([label])[0])  # 这里需要使用标签编码器进行编码

val_pre_auc = np.array(val_pre_auc)
val_label_auc = np.array(val_label_auc)

auc_score_val = [metrics.roc_auc_score(val_label_auc[:, i], val_pre_auc[:, i]) for i in range(n_classes)]


train_pre_auc   = []
train_label_auc = []

for images, labels in train_ds:
    for image, label in zip(images, labels):
        img_array_train = tf.expand_dims(image, 0) 
        prediction_auc = model.predict(img_array_train)
        train_pre_auc.append(prediction_auc[0])
        train_label_auc.append(lb.transform([label])[0])

train_pre_auc = np.array(train_pre_auc)
train_label_auc = np.array(train_label_auc)

auc_score_train = [metrics.roc_auc_score(train_label_auc[:, i], train_pre_auc[:, i]) for i in range(n_classes)]

fig, axs = plt.subplots(n_classes, figsize=(5, 20))

for i in range(n_classes):
    plot_roc(axs[i], 'validation AUC for class {0}: {1:.4f}'.format(i, auc_score_val[i]), val_label_auc[:, i] , val_pre_auc[:, i], color="red", linestyle='--')
    plot_roc(axs[i], 'training AUC for class {0}: {1:.4f}'.format(i, auc_score_train[i]), train_label_auc[:, i], train_pre_auc[:, i], color="blue", linestyle='--')
    axs[i].legend(loc='lower right')

plt.tight_layout()
plt.savefig("roc.pdf", dpi=300,format="pdf")

for i in range(n_classes):
    print("Class {0} 训练集的AUC值为：".format(i), auc_score_train[i], "验证集的AUC值为：", auc_score_val[i])

（b）调教GPT-4的过程

（b1） 咒语：请根据{代码1}，改写和续写《代码2》。代码1：{也就是之前用tensorflow写的误判病例分析部分}；代码2：《也就是修改之前的Mobilenet模型建模代码》

然后根据具体情况调整即可，当然是在GPT的帮助下。

三、输出结果

（1） 学习曲线

（2）混淆矩阵

（3）性能参数

（4）ROC曲线

（4.1）和在一起的：

（4.2）分开的：

四、数据

链接：https://pan.baidu.com/s/1rqu15KAUxjNBaWYfEmPwgQ?pwd=xfyn

提取码：xfyn

相关推荐
阿_旭4 小时前
一文读懂| 自注意力与交叉注意力机制在计算机视觉中作用与基本原理
人工智能·深度学习·计算机视觉·cross-attention·self-attention
王哈哈^_^4 小时前
【数据集】【YOLO】【目标检测】交通事故识别数据集 8939 张，YOLO道路事故目标检测实战训练教程！
前端·人工智能·深度学习·yolo·目标检测·计算机视觉·pyqt
Power20246665 小时前
NLP论文速读|LongReward:基于AI反馈来提升长上下文大语言模型
人工智能·深度学习·机器学习·自然语言处理·nlp
数据猎手小k5 小时前
AIDOVECL数据集：包含超过15000张AI生成的车辆图像数据集，目的解决旨在解决眼水平分类和定位问题。
人工智能·分类·数据挖掘
YRr YRr5 小时前
深度学习：循环神经网络（RNN）详解
人工智能·rnn·深度学习
sp_fyf_20246 小时前
计算机前沿技术-人工智能算法-大语言模型-最新研究进展-2024-11-01
人工智能·深度学习·神经网络·算法·机器学习·语言模型·数据挖掘
红客5976 小时前
Transformer和BERT的区别
深度学习·bert·transformer
多吃轻食6 小时前
大模型微调技术 --＞ 脉络
人工智能·深度学习·神经网络·自然语言处理·embedding
charles_vaez6 小时前
开源模型应用落地-glm模型小试-glm-4-9b-chat-快速体验（一）
深度学习·语言模型·自然语言处理
YRr YRr6 小时前
深度学习：Transformer Decoder详解
人工智能·深度学习·transformer
热门推荐
01如何才能让手机厂商主动拥抱华为，接入鸿蒙系统？02【HarmonyOS】HUAWEI DevEco Studio 下载地址汇总03组基轨迹建模 GBTM的介绍与实现（Stata 或 R）04【TC3xx芯片】TC3xx芯片电源管理系统PMS详解05基于YOLOv10深度学习的CT扫描图像肾结石智能检测系统【python源码+Pyqt5界面+数据集+训练代码】深度学习实战、目标检测06【经验分享】Ubuntu22.04安装微信(linux官方版)07Macbook pro M1 安装Ubuntu教程08RAG 实践- Ollama+RagFlow 部署本地知识库09Windows10安装PCL1.14.0及点云配准10文件或文件夹名称中有空格如何批量去除