CNN识别围棋棋盘240814

​
import tensorflow as tf
from tensorflow.keras import layers, models
from tensorflow.keras.datasets import mnist

# 假设输入图像的形状是19x19x1（灰度图像）
input_shape = (19, 19, 1)
num_classes = 10  # 假设有10个类别

# 创建模型
model = models.Sequential()

# 添加输入层
model.add(layers.Input(shape=input_shape))

# 添加卷积层
model.add(layers.Conv2D(32, (3, 3), activation='relu', padding='same'))
model.add(layers.MaxPooling2D((2, 2)))

# 添加更多卷积层
model.add(layers.Conv2D(64, (3, 3), activation='relu', padding='same'))
model.add(layers.MaxPooling2D((2, 2)))

# 添加全连接层
model.add(layers.Flatten())
model.add(layers.Dense(128, activation='relu'))

# 添加输出层，输出形状为 (19, 19, num_classes)
model.add(layers.Dense(19 * 19 * num_classes, activation='softmax'))
model.add(layers.Reshape((19, 19, num_classes)))

# 编译模型，使用 categorical_crossentropy 作为损失函数
model.compile(optimizer='adam',
              loss='categorical_crossentropy',
              metrics=['accuracy'])

# 加载示例数据集
# 为了演示，这里使用 MNIST 数据集，并将其重新调整为 19x19 的形状
(train_data, train_labels), (test_data, test_labels) = mnist.load_data()

# 调整数据形状为 (19, 19, 1) 并归一化
train_data = tf.image.resize(train_data[..., tf.newaxis], [19, 19]).numpy() / 255.0
test_data = tf.image.resize(test_data[..., tf.newaxis], [19, 19]).numpy() / 255.0

# 将标签数据转换为 one-hot 编码形式
train_labels = tf.keras.utils.to_categorical(train_labels, num_classes)
test_labels = tf.keras.utils.to_categorical(test_labels, num_classes)

# 训练模型
model.fit(train_data, train_labels, epochs=5, batch_size=32, validation_data=(test_data, test_labels))

# 模型评估
test_loss, test_acc = model.evaluate(test_data, test_labels)
print(f"Test accuracy: {test_acc}")

​