机器学习—飞行器类型图片识别

# 导入所需要的库
import pandas as pd
import numpy as np
import tensorflow as tf
import matplotlib.pyplot as plt
import matplotlib.font_manager as fm
from keras.preprocessing.image import ImageDataGenerator
from sklearn.model_selection import train_test_split
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Conv2D, Activation, MaxPool2D, Flatten, Dense, Dropout
from tensorflow.keras.preprocessing.image import ImageDataGenerator
from tensorflow.keras import optimizers
from pathlib import Path
from matplotlib import font_manager
# 设置全局字体
my_font=font_manager.FontProperties(fname="C:\Windows\Fonts\STSONG.TTF")
plt.rcParams['font.family'] = 'STSong'  # 替换成所选的中文字体

# 读取6种飞行器类型的图片路径和标签
filepaths = []
labels = []
categories = ['drone', 'fighterjet', 'helicopter', 'missile', 'passengerplane', 'rocket']
for category in categories:
    dir_path = Path('E:/python/data/planes/' + category)
    filepaths_category = list(dir_path.glob('**/*.JPG'))
    filepaths.extend(filepaths_category)
    labels.extend([category] * len(filepaths_category))

#图像数据处理
categories = ['drone', 'fighterjet', 'helicopter', 'missile', 'passengerplane', 'rocket']
filepaths = []
labels = []
for category in categories:
    dir_path = Path('E:/python/data/planes/' + category)
    filepaths_category = list(dir_path.glob('**/*.JPG'))
    filepaths.extend(filepaths_category)
    labels.extend([category] * len(filepaths_category))
data = pd.DataFrame({'FilePaths': filepaths, 'labels': labels})
print(len(data['labels']))
print(len(data))
print(data.head())
print(data.info())

from sklearn.model_selection import train_test_split

data['FilePaths'] = data['FilePaths'].astype(str)

#9:1划分为训练集和测试集
X_train_test, X_test = train_test_split(data, test_size=0.1, stratify=data['labels'])
print('测试集形状', X_test.shape)

# 4:1划分为训练集和验证集
X_train, X_val = train_test_split(X_train_test, test_size=0.2, stratify=X_train_test['labels'])
print('训练集形状', X_train.shape)
print('验证集形状', X_val.shape)

# 查看各类型的图片张数
print(X_train['labels'].value_counts())
print(X_train['FilePaths'].shape)

from tensorflow.keras.preprocessing.image import ImageDataGenerator

img_preprocessing = ImageDataGenerator(rescale=1./255)

data_gen_params = {
    'target_size': (112, 112),
    'color_mode': 'rgb',
    'class_mode': 'categorical',
    'batch_size': 32,
    'seed': 30
}

x_train = img_preprocessing.flow_from_dataframe(dataframe=X_train, x_col='FilePaths', y_col='labels', **data_gen_params)
x_test = img_preprocessing.flow_from_dataframe(dataframe=X_test, x_col='FilePaths', y_col='labels', **data_gen_params)
x_val = img_preprocessing.flow_from_dataframe(dataframe=X_val, x_col='FilePaths', y_col='labels', **data_gen_params)

from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Conv2D, Activation, MaxPool2D, Flatten, Dense, Dropout
from tensorflow.keras import optimizers

# 构建神经网络模型
model = Sequential()

# Conv2D层，32个滤波器
model.add(Conv2D(filters=32, kernel_size=(3, 3), padding='same'))
model.add(Activation('relu'))
model.add(MaxPool2D(pool_size=(2, 2), strides=2, padding='valid'))

# Conv2D层，64个滤波器
model.add(Conv2D(filters=64, kernel_size=(3, 3), padding='same'))
model.add(Activation('relu'))
model.add(MaxPool2D(pool_size=(2, 2), strides=2, padding='valid'))

# Conv2D层，128个滤波器
model.add(Conv2D(filters=128, kernel_size=(3, 3), padding='same'))
model.add(Activation('relu'))
model.add(MaxPool2D(pool_size=(2, 2), strides=2, padding='valid'))

# Conv2D层，256个滤波器
model.add(Conv2D(filters=256, kernel_size=(3, 3), padding='same'))
model.add(Activation('relu'))
model.add(MaxPool2D(pool_size=(2, 2), strides=2, padding='valid'))

# Conv2D层，512个滤波器
model.add(Conv2D(filters=512, kernel_size=(3, 3), padding='same'))
model.add(Activation('relu'))
model.add(MaxPool2D(pool_size=(2, 2), strides=2, padding='valid'))

# Conv2D层，1024个滤波器
model.add(Conv2D(filters=1024, kernel_size=(3, 3), padding='same'))
model.add(Activation('relu'))
model.add(MaxPool2D(pool_size=(2, 2), strides=2, padding='valid'))

# 展平数据，降维
model.add(Flatten())

# 全连接层
model.add(Dense(2048))
model.add(Activation('relu'))

# 减少过拟合
model.add(Dropout(0.5))

# 全连接层
model.add(Dense(6))  # 识别6种类别
model.add(Activation('softmax'))  # 使用softmax进行分类

# 模型编译
model.compile(optimizer=optimizers.RMSprop(learning_rate=1e-4),
              loss="categorical_crossentropy",
              metrics=["accuracy"])

print('————————开始训练————————')
print('模型训练中——————————————')
model1 = model.fit(x_train, validation_data=x_val, epochs=500)
model.summary()

print('\n验证中-----')
loss, accuracy = model.evaluate(x_val)
print('\n验证损失:', loss)
print('\n验证准确率:', accuracy)

# 保存模型
model.save('E:\python\data\model1')

import matplotlib.pyplot as plt
# 绘制model1模型损失变化曲线
loss = model1.history["loss"]
val_loss = model1.history["val_loss"]
plt.plot(loss, "bo-", label="训练损失")
plt.plot(val_loss, "ro--", label="验证损失")
plt.title("训练和验证损失")
plt.xlabel("训练迭代次数")
plt.ylabel("损失")
plt.legend()
plt.savefig('model1模型损失变化曲线图.jpg')
plt.show()

# 绘制model1模型准确率变化曲线
acc = model1.history["accuracy"]
val_acc = model1.history["val_accuracy"]
plt.plot(acc, "bo-", label="训练准确率")
plt.plot(val_acc, "ro--", label="验证准确率")
plt.title("训练和验证准确率")
plt.xlabel("训练迭代次数")
plt.ylabel("准确率")
plt.legend()
plt.savefig('model1模型准确率变化曲线图.jpg')
plt.show()

from keras.preprocessing.image import ImageDataGenerator

# 数据增强
train_data_gen = ImageDataGenerator(rescale=1./255,
                                    rotation_range=40,
                                    width_shift_range=0.2,
                                    height_shift_range=0.2,
                                    shear_range=0.2,
                                    zoom_range=0.2,
                                    horizontal_flip=True,
                                    fill_mode='nearest')

val_data_gen = ImageDataGenerator(rescale=1./255)

train_generator = train_data_gen.flow_from_dataframe(dataframe=X_train,
                                                     x_col='FilePaths',
                                                     y_col='labels',
                                                     target_size=(112, 112),
                                                     color_mode='rgb',
                                                     class_mode='categorical',
                                                     batch_size=32,
                                                     seed=30)


val_generator = val_data_gen.flow_from_dataframe(dataframe=X_val,
                                                 x_col='FilePaths',
                                                 y_col='labels',
                                                 target_size=(112, 112),
                                                 color_mode='rgb',
                                                 class_mode='categorical',
                                                 batch_size=32,
                                                 seed=30)

print('——————开始训练——————')
print('模型训练中————————')
model2 = model.fit(
x_train, validation_data=x_val, epochs=300)
model.summary()

print('\n验证中-----')
loss, accuracy = model.evaluate(x_val)
print('\n验证损失:', loss)
print('\n验证准确率:', accuracy)

# 保存模型
model.save('E:\python\data\model2')

# 绘制model2模型损失变化曲线
loss = model2.history["loss"]
val_loss = model2.history["val_loss"]
plt.plot(loss, "bo-", label="训练损失")
plt.plot(val_loss, "ro--", label="验证损失")
plt.title("训练和验证损失")
plt.xlabel("训练迭代次数")
plt.ylabel("损失")
plt.legend()
plt.savefig('model2模型损失变化曲线图.jpg')
plt.show()

# 绘制model2模型准确率变化曲线
acc = model2.history["accuracy"]
val_acc = model2.history["val_accuracy"]
plt.plot(acc, "bo-", label="训练准确率")
plt.plot(val_acc, "ro--", label="验证准确率")
plt.title("训练和验证准确率")
plt.xlabel("训练迭代次数")
plt.ylabel("准确率")
plt.legend()
plt.savefig('model2模型准确率变化曲线图.jpg')
plt.show()

from keras.models import load_model
import numpy as np
import matplotlib.pyplot as plt

# 加载model1模型
model1 = load_model('E:\python\data\model1')
l1 = ['drone', 'fighterjet', 'helicopter', 'missile', 'passengerplane', 'rocket']

# model1模型预测
i = 10
img1, label1 = x_test.next()

plt.figure(figsize=(12, 7))
plt.subplot(2, 1, 1)
plt.imshow((img1[i] * 255).astype('uint8'))
plt.title('实际类型为：' + l1[np.argmax(label1[i])])

x_test_img1 = img1[i].reshape(1, 112, 112, 3).astype("float32")
pr1 = model1.predict(x_test_img1)

plt.subplot(2, 1, 2)
plt.title("预测结果的概率饼图")
plt.pie(pr1.reshape(6), labels=l1, autopct='%1.1f%%')
plt.axis('equal')
plt.show()

print('实际类型：' + l1[np.argmax(label1[i])])
print('model1模型预测的结果是：' + l1[np.argmax(pr1)])

# 加载model2模型
model2 = load_model('E:\python\data\model2')

# model2模型预测
i = 10
img2, label2 = x_test.next()
l2 = ['drone', 'fighterjet', 'helicopter', 'missile', 'passengerplane', 'rocket']
plt.figure(figsize=(12, 7))
plt.subplot(2, 1, 1)
plt.imshow((img2[i] * 255).astype('uint8'))
plt.title('实际类型为：' + l2[np.argmax(label2[i])])

x_test_img2 = img2[i].reshape(1, 112, 112, 3).astype("float32")
pr2 = model2.predict(x_test_img2)

plt.subplot(2, 1, 2) 
plt.title("预测结果的概率饼图")
plt.pie(pr2.reshape(6), labels=l2, autopct='%1.1f%%')
plt.axis('equal')
plt.show()

print('实际类型：' + l2[np.argmax(label2[i])])
print('model2模型预测的结果是：' + l2[np.argmax(pr2)])