keras实现用Googlenet对四种天气分类，并绘制图像

首先，我们需要准备数据集。假设我们有一个包含四种天气（晴天、多云、雨天、雪天）的数据集，其中每个类别包含100张图片。我们将这些图片分为训练集和测试集，其中训练集包含80张图片，测试集包含20张图片。

接下来，我们需要使用Keras实现GoogLeNet模型。以下是代码：

from keras.layers import Input, Conv2D, MaxPooling2D, Dropout, Dense, AveragePooling2D, Flatten, concatenate
from keras.models import Model

def inception_module(x, filters):
    filter1, filter3, filter5, filterpool = filters
    
    conv1 = Conv2D(filter1, (1,1), padding='same', activation='relu')(x)
    
    conv3 = Conv2D(filter3, (1,1), padding='same', activation='relu')(x)
    conv3 = Conv2D(filter3, (3,3), padding='same', activation='relu')(conv3)
    
    conv5 = Conv2D(filter5, (1,1), padding='same', activation='relu')(x)
    conv5 = Conv2D(filter5, (5,5), padding='same', activation='relu')(conv5)
    
    maxpool = MaxPooling2D((3,3), strides=(1,1), padding='same')(x)
    maxpool = Conv2D(filterpool, (1,1), padding='same', activation='relu')(maxpool)
    
    output = concatenate([conv1, conv3, conv5, maxpool], axis=3)
    
    return output

input_layer = Input(shape=(224,224,3))

conv1 = Conv2D(64, (7,7), strides=(2,2), padding='same', activation='relu')(input_layer)
maxpool1 = MaxPooling2D((3,3), strides=(2,2), padding='same')(conv1)

conv2 = Conv2D(64, (1,1), padding='same', activation='relu')(maxpool1)
conv2 = Conv2D(192, (3,3), padding='same', activation='relu')(conv2)
maxpool2 = MaxPooling2D((3,3), strides=(2,2), padding='same')(conv2)

inception3a = inception_module(maxpool2, [64, 96, 128, 16])
inception3b = inception_module(inception3a, [128, 128, 192, 32])
maxpool3 = MaxPooling2D((3,3), strides=(2,2), padding='same')(inception3b)

inception4a = inception_module(maxpool3, [192, 96, 208, 16])
inception4b = inception_module(inception4a, [160, 112, 224, 24])
inception4c = inception_module(inception4b, [128, 128, 256, 24])
inception4d = inception_module(inception4c, [112, 144, 288, 32])
inception4e = inception_module(inception4d, [256, 160, 320, 32])
maxpool4 = MaxPooling2D((3,3), strides=(2,2), padding='same')(inception4e)

inception5a = inception_module(maxpool4, [256, 160, 320, 32])
inception5b = inception_module(inception5a, [384, 192, 384, 48])
averagepool = AveragePooling2D((7,7), strides=(7,7))(inception5b)

flatten = Flatten()(averagepool)
dropout = Dropout(0.4)(flatten)
output_layer = Dense(4, activation='softmax')(dropout)

model = Model(inputs=input_layer, outputs=output_layer)
model.summary()

然后，我们需要编译并训练模型。以下是代码：

from keras.optimizers import SGD
from keras.preprocessing.image import ImageDataGenerator

train_datagen = ImageDataGenerator(rescale=1./255)
test_datagen = ImageDataGenerator(rescale=1./255)

train_generator = train_datagen.flow_from_directory(
    'train/',
    target_size=(224,224),
    batch_size=16,
    class_mode='categorical'
)

test_generator = test_datagen.flow_from_directory(
    'test/',
    target_size=(224,224),
    batch_size=16,
    class_mode='categorical'
)

sgd = SGD(lr=0.01, decay=1e-6, momentum=0.9, nesterov=True)
model.compile(optimizer=sgd, loss='categorical_crossentropy', metrics=['accuracy'])

model.fit_generator(train_generator, steps_per_epoch=80, epochs=50, validation_data=test_generator, validation_steps=20)

最后，我们可以使用以下代码进行预测并绘制图像：

import numpy as np
from keras.preprocessing import image
import matplotlib.pyplot as plt

test_image = image.load_img('test_image.jpg', target_size=(224,224))
test_image = image.img_to_array(test_image)
test_image = np.expand_dims(test_image, axis=0)
result = model.predict(test_image)

labels = ['Sunny', 'Cloudy', 'Rainy', 'Snowy']

plt.imshow(image.load_img('test_image.jpg'))
plt.title(labels[np.argmax(result)])
plt.show()

注意，上述代码中的`'train/'`和`'test/'`分别表示训练集和测试集的文件夹路径，你需要根据实际情况修改这些路径。另外，`'test_image.jpg'`表示要进行预测的图像文件名，你也需要根据实际情况修改这个文件名。