基于Python实现地标景点识别功能

import cv2
import os

# 数据预处理函数
def preprocess_data(input_dir, output_dir):
    # 创建输出目录
    if not os.path.exists(output_dir):
        os.makedirs(output_dir)

    # 获取输入目录下的所有图片文件
    image_files = [f for f in os.listdir(input_dir) if f.endswith('.jpg') or f.endswith('.png')]

    for image_file in image_files:
        # 读取图像
        image_path = os.path.join(input_dir, image_file)
        image = cv2.imread(image_path)

        # 调整图像大小为统一尺寸
        resized_image = cv2.resize(image, (224, 224))

        # 归一化处理
        normalized_image = resized_image / 255.0

        # 图像色彩空间转换为灰度图像
        gray_image = cv2.cvtColor(normalized_image, cv2.COLOR_BGR2GRAY)

        # 保存预处理后的图像
        output_path = os.path.join(output_dir, image_file)
        cv2.imwrite(output_path, gray_image)

    print("数据预处理完成！")

# 输入目录和输出目录
input_directory = "input_data"
output_directory = "preprocessed_data"

# 运行数据预处理函数
preprocess_data(input_directory, output_directory)

import tensorflow as tf
from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import Conv2D, MaxPooling2D, Flatten, Dense

# 定义卷积神经网络模型
def build_cnn_model(input_shape, num_classes):
    model = Sequential()

    # 添加卷积层和池化层
    model.add(Conv2D(32, (3, 3), activation='relu', input_shape=input_shape))
    model.add(MaxPooling2D((2, 2)))

    model.add(Conv2D(64, (3, 3), activation='relu'))
    model.add(MaxPooling2D((2, 2)))

    model.add(Conv2D(128, (3, 3), activation='relu'))
    model.add(MaxPooling2D((2, 2)))

    model.add(Flatten())

    # 添加全连接层
    model.add(Dense(512, activation='relu'))
    model.add(Dense(num_classes, activation='softmax'))

    return model

# 输入图像的大小和类别数
input_shape = (224, 224, 1)
num_classes = 10

# 构建CNN模型
model = build_cnn_model(input_shape, num_classes)

# 编译模型
model.compile(optimizer='adam',
              loss=tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True),
              metrics=['accuracy'])

# 输出模型概述
model.summary()

import tensorflow as tf
from tensorflow.keras.preprocessing.image import ImageDataGenerator
from tensorflow.keras.callbacks import ModelCheckpoint

# 数据路径和参数设置
train_data_dir = 'train_data'
validation_data_dir = 'validation_data'
input_shape = (224, 224, 1)
batch_size = 32
epochs = 20

# 数据增强
train_datagen = ImageDataGenerator(
    rescale=1./255,
    shear_range=0.2,
    zoom_range=0.2,
    horizontal_flip=True)

validation_datagen = ImageDataGenerator(rescale=1./255)

# 读取数据
train_generator = train_datagen.flow_from_directory(
    train_data_dir,
    target_size=(input_shape[0], input_shape[1]),
    batch_size=batch_size,
    color_mode='grayscale',
    class_mode='sparse')

validation_generator = validation_datagen.flow_from_directory(
    validation_data_dir,
    target_size=(input_shape[0], input_shape[1]),
    batch_size=batch_size,
    color_mode='grayscale',
    class_mode='sparse')

# 构建CNN模型
model = build_cnn_model(input_shape, num_classes)

# 编译模型
model.compile(optimizer='adam',
              loss=tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True),
              metrics=['accuracy'])

# 回调函数：保存最佳模型
checkpoint_callback = ModelCheckpoint('best_model.h5', monitor='val_accuracy',
                                       save_best_only=True, mode='max', verbose=1)

# 训练模型
history = model.fit(train_generator,
                    epochs=epochs,
                    steps_per_epoch=train_generator.samples // batch_size,
                    validation_data=validation_generator,
                    validation_steps=validation_generator.samples // batch_size,
                    callbacks=[checkpoint_callback])

# 保存模型
model.save('final_model.h5')

# 输出训练结果
print("训练准确率：", history.history['accuracy'][-1])
print("验证准确率：", history.history['val_accuracy'][-1])