Keras上的Conv2d尺寸错误

import numpy as np
from tensorflow.keras.layers import Conv2D, Input
from tensorflow.keras.models import Model

# 示例输入数据
input_data = np.random.rand(32, 64, 64, 1)  # 32张64x64的灰度图像

# 定义输入层
input_layer = Input(shape=(64, 64, 1))

# 定义Conv2D层
conv_layer = Conv2D(filters=32, kernel_size=(3, 3), activation='relu')(input_layer)

# 构建模型
model = Model(inputs=input_layer, outputs=conv_layer)

# 打印模型摘要
model.summary()

from tensorflow.keras.preprocessing.image import ImageDataGenerator

# 示例数据生成器
datagen = ImageDataGenerator(rescale=1./255, validation_split=0.2)

# 加载并预处理数据
train_generator = datagen.flow_from_directory(
    'path_to_train_data',
    target_size=(64, 64),
    batch_size=32,
    class_mode='binary',
    subset='training'
)

validation_generator = datagen.flow_from_directory(
    'path_to_train_data',
    target_size=(64, 64),
    batch_size=32,
    class_mode='binary',
    subset='validation'
)

input_layer = Input(shape=(64, 64, 1))  # 注意这里的shape
conv_layer = Conv2D(filters=32, kernel_size=(3, 3), activation='relu')(input_layer)

import numpy as np
from tensorflow.keras.layers import Conv2D, Input, Flatten, Dense
from tensorflow.keras.models import Model

# 示例输入数据
input_data = np.random.rand(32, 64, 64, 1)  # 32张64x64的灰度图像

# 定义输入层
input_layer = Input(shape=(64, 64, 1))

# 定义Conv2D层
conv_layer = Conv2D(filters=32, kernel_size=(3, 3), activation='relu')(input_layer)
flatten_layer = Flatten()(conv_layer)
dense_layer = Dense(units=10, activation='softmax')(flatten_layer)

# 构建模型
model = Model(inputs=input_layer, outputs=dense_layer)

# 编译模型
model.compile(optimizer='adam', loss='sparse_categorical_crossentropy', metrics=['accuracy'])

# 打印模型摘要
model.summary()

# 训练模型
model.fit(input_data, np.random.randint(0, 10, size=(32,)), epochs=5)