15.Tensorflow2.0 Keras高层接口

1. Keras.Metrics （度量指标）

1.1. Build a meter

1.2. Update data

1.3. Get Average data

1.4. Clear buffer

1.5. Code

import tensorflow as tf
from tensorflow.keras import datasets, layers, optimizers, Sequential, metrics


def pre_process(x, y):
    x = tf.cast(x, dtype=tf.float32) / 255.
    y = tf.cast(y, dtype=tf.int32)

    return x, y


batch_size = 128
(x, y), (x_val, y_val) = datasets.mnist.load_data()
print('datasets: ', x.shape, y.shape, x.min(), x.max())

db = tf.data.Dataset.from_tensor_slices((x, y))
db = db.map(pre_process).shuffle(60000).batch(batch_size).repeat(10)

ds_val = tf.data.Dataset.from_tensor_slices((x_val, y_val))
ds_val = ds_val.map(pre_process).batch(batch_size)

model = Sequential([layers.Dense(256, activation='relu'),
                    layers.Dense(128, activation='relu'),
                    layers.Dense(64, activation='relu'),
                    layers.Dense(32, activation='relu'),
                    layers.Dense(10, activation='relu')
                    ])

model.build(input_shape=(None, 28 * 28))
model.summary()

optimizer = optimizers.Adam(learning_rate=1e-3)

acc_meter = metrics.Accuracy()
loss_meter = metrics.Mean()

for step, (x, y) in enumerate(db):
    with tf.GradientTape() as tape:
        # [b, 28, 28] => [b, 784]
        x = tf.reshape(x, (-1, 28 * 28))
        # [b, 784] => [b, 10]
        out = model(x)
        # [b] => [b, 10]
        y_onehot = tf.one_hot(y, depth=10)
        # [b]
        loss = tf.reduce_mean(tf.losses.categorical_crossentropy(y_onehot, out, from_logits=True))
        loss_meter.update_state(loss)

    grads = tape.gradient(loss, model.trainable_variables)
    optimizer.apply_gradients(zip(grads, model.trainable_variables))

    if step % 100 == 0:
        print(step, 'loss:', loss_meter.result().numpy())

    # evaluate
    if step % 500 == 0:
        total, total_correct = 0., 0
        acc_meter.reset_states()

        for step, (x, y) in enumerate(ds_val):
            # [b, 28, 28] => [b, 784]
            x = tf.reshape(x, (-1, 28*28))
            # [b, 784] => [b, 10]
            out = model(x)

            # [b, 10] => [b]
            pred = tf.argmax(out, axis=1)
            pred = tf.cast(pred, dtype=tf.int32)
            # bool type
            correct = tf.equal(pred, y)
            # bool tensor => int tensor => numpy
            total_correct += tf.reduce_sum(tf.cast(correct, dtype=tf.int32)).numpy()
            total += x.shape[0]

            acc_meter.update_state(y, pred)

        print(step, " Evaluate acc:", total_correct/total, acc_meter.result().numpy())

1.6. 测试结果

0 loss: 2.2957244
78  Evaluate acc: 0.1831 0.1831
100 loss: 1.2260486
200 loss: 1.0626991
300 loss: 0.9883962
400 loss: 0.9474341
500 loss: 0.92040706
78  Evaluate acc: 0.6759 0.6759
600 loss: 0.8984385
700 loss: 0.88047
800 loss: 0.86719453
900 loss: 0.85594875
1000 loss: 0.8451226
78  Evaluate acc: 0.6831 0.6831
1100 loss: 0.83795416
1200 loss: 0.8301951
1300 loss: 0.8243892
1400 loss: 0.81990254
1500 loss: 0.8136047
78  Evaluate acc: 0.6837 0.6837
1600 loss: 0.80968654
1700 loss: 0.80532765
1800 loss: 0.8013762
1900 loss: 0.79705083
2000 loss: 0.792495
78  Evaluate acc: 0.6869 0.6869
2100 loss: 0.78891665
2200 loss: 0.78668064
2300 loss: 0.7847733
2400 loss: 0.7820607
2500 loss: 0.7791471
78  Evaluate acc: 0.6842 0.6842
2600 loss: 0.77668834
2700 loss: 0.77496046
2800 loss: 0.7732897
2900 loss: 0.7712144
3000 loss: 0.7693945
78  Evaluate acc: 0.6859 0.6859
3100 loss: 0.7674685
3200 loss: 0.7662053
3300 loss: 0.7646868
3400 loss: 0.7630929
3500 loss: 0.7568395
78  Evaluate acc: 0.7754 0.7754
3600 loss: 0.749667
3700 loss: 0.74337995
3800 loss: 0.73666996
3900 loss: 0.73019713
4000 loss: 0.7240193
78  Evaluate acc: 0.7788 0.7788
4100 loss: 0.7184695
4200 loss: 0.7131745
4300 loss: 0.70784307
4400 loss: 0.70276886
4500 loss: 0.6979761
78  Evaluate acc: 0.7794 0.7794
4600 loss: 0.69365555

2. Compile&Fit

2.1. Code

import tensorflow as tf
from tensorflow.keras import datasets, layers, optimizers, Sequential


def pre_process(x, y):
    """
    x is a simple image, not a batch
    """
    x = tf.cast(x, dtype=tf.float32) / 255.
    x = tf.reshape(x, [28 * 28])
    y = tf.cast(y, dtype=tf.int32)
    y = tf.one_hot(y, depth=10)
    return x, y


batch_size = 128
(x, y), (x_val, y_val) = datasets.mnist.load_data()
print('datasets:', x.shape, y.shape, x.min(), x.max())

db = tf.data.Dataset.from_tensor_slices((x, y))
db = db.map(pre_process).shuffle(60000).batch(batch_size)
ds_val = tf.data.Dataset.from_tensor_slices((x_val, y_val))
ds_val = ds_val.map(pre_process).batch(batch_size)


network = Sequential([layers.Dense(256, activation='relu'),
                      layers.Dense(128, activation='relu'),
                      layers.Dense(64, activation='relu'),
                      layers.Dense(32, activation='relu'),
                      layers.Dense(10)])
network.build(input_shape=(None, 28 * 28))
network.summary()

network.compile(optimizer=optimizers.Adam(lr=0.01),
                loss=tf.losses.CategoricalCrossentropy(from_logits=True),
                metrics=['accuracy']
                )

network.fit(db, epochs=5, validation_data=ds_val, validation_freq=2)

network.evaluate(ds_val)

2.2. 测试结果

3. 自定义网络

import tensorflow as tf
from tensorflow import keras
from tensorflow.keras import datasets, layers, optimizers, Sequential


def pre_process(x, y):
    """
    x is a simple image, not a batch
    """
    x = tf.cast(x, dtype=tf.float32) / 255.
    x = tf.reshape(x, [28 * 28])
    y = tf.cast(y, dtype=tf.int32)
    y = tf.one_hot(y, depth=10)
    return x, y


batch_size = 128
(x, y), (x_val, y_val) = datasets.mnist.load_data()
print('datasets:', x.shape, y.shape, x.min(), x.max())

db = tf.data.Dataset.from_tensor_slices((x, y))
db = db.map(pre_process).shuffle(60000).batch(batch_size)
ds_val = tf.data.Dataset.from_tensor_slices((x_val, y_val))
ds_val = ds_val.map(pre_process).batch(batch_size)

network = Sequential([layers.Dense(256, activation='relu'),
                      layers.Dense(128, activation='relu'),
                      layers.Dense(64, activation='relu'),
                      layers.Dense(32, activation='relu'),
                      layers.Dense(10)])
network.build(input_shape=(None, 28 * 28))
network.summary()


class MyDense(layers.Layer):

    def __init__(self, inp_dim, outp_dim):
        super(MyDense, self).__init__()

        self.kernel = self.add_weight('w', [inp_dim, outp_dim])
        self.bias = self.add_weight('b', [outp_dim])

    def call(self, inputs, training=None):
        out = inputs @ self.kernel + self.bias

        return out


class MyModel(keras.Model):

    def __init__(self):
        super(MyModel, self).__init__()

        self.fc1 = MyDense(28 * 28, 256)
        self.fc2 = MyDense(256, 128)
        self.fc3 = MyDense(128, 64)
        self.fc4 = MyDense(64, 32)
        self.fc5 = MyDense(32, 10)

    def call(self, inputs, training=None):
        x = self.fc1(inputs)
        x = tf.nn.relu(x)
        x = self.fc2(x)
        x = tf.nn.relu(x)
        x = self.fc3(x)
        x = tf.nn.relu(x)
        x = self.fc4(x)
        x = tf.nn.relu(x)
        x = self.fc5(x)

        return x


network = MyModel()

network.compile(optimizer=optimizers.Adam(lr=1e-3),
                loss=tf.losses.CategoricalCrossentropy(from_logits=True),
                metrics=['accuracy']
                )

network.fit(db, epochs=5, validation_data=ds_val,
            validation_freq=2)

network.evaluate(ds_val)

4. 模型的保存与加载

4.1. save/load weights

import tensorflow as tf
from tensorflow.keras import datasets, layers, optimizers, Sequential, metrics


def pre_process(x, y):
    """
    x is a simple image, not a batch
    """
    x = tf.cast(x, dtype=tf.float32) / 255.
    x = tf.reshape(x, [28 * 28])
    y = tf.cast(y, dtype=tf.int32)
    y = tf.one_hot(y, depth=10)
    return x, y


batch_size = 128
(x, y), (x_val, y_val) = datasets.mnist.load_data()
print('datasets:', x.shape, y.shape, x.min(), x.max())

db = tf.data.Dataset.from_tensor_slices((x, y))
db = db.map(pre_process).shuffle(60000).batch(batch_size)
ds_val = tf.data.Dataset.from_tensor_slices((x_val, y_val))
ds_val = ds_val.map(pre_process).batch(batch_size)

network = Sequential([layers.Dense(256, activation='relu'),
                      layers.Dense(128, activation='relu'),
                      layers.Dense(64, activation='relu'),
                      layers.Dense(32, activation='relu'),
                      layers.Dense(10)])

network.build(input_shape=(None, 28 * 28))
network.summary()

network.compile(optimizer=optimizers.Adam(lr=0.01),
                loss=tf.losses.CategoricalCrossentropy(from_logits=True),
                metrics=['accuracy']
                )

network.fit(db, epochs=3, validation_data=ds_val, validation_freq=2)

network.evaluate(ds_val)
# 保存权重， 参数
network.save_weights('ckpt/weights.ckpt')
print('saved weights.')
del network

network = Sequential([layers.Dense(256, activation='relu'),
                      layers.Dense(128, activation='relu'),
                      layers.Dense(64, activation='relu'),
                      layers.Dense(32, activation='relu'),
                      layers.Dense(10)])
network.compile(optimizer=optimizers.Adam(lr=0.01),
                loss=tf.losses.CategoricalCrossentropy(from_logits=True),
                metrics=['accuracy']
                )
# 加载权重， 参数
network.load_weights('ckpt/weights.ckpt')
print('loaded weights!')
# 计算准确率
network.evaluate(ds_val)

4.2. save/load entire model

import tensorflow as tf
from tensorflow.keras import datasets, layers, optimizers, Sequential, metrics


def pre_process(x, y):
    """
    x is a simple image, not a batch
    """
    x = tf.cast(x, dtype=tf.float32) / 255.
    x = tf.reshape(x, [28 * 28])
    y = tf.cast(y, dtype=tf.int32)
    y = tf.one_hot(y, depth=10)
    return x, y


batch_size = 128
(x, y), (x_val, y_val) = datasets.mnist.load_data()
print('datasets:', x.shape, y.shape, x.min(), x.max())

db = tf.data.Dataset.from_tensor_slices((x, y))
db = db.map(pre_process).shuffle(60000).batch(batch_size)
ds_val = tf.data.Dataset.from_tensor_slices((x_val, y_val))
ds_val = ds_val.map(pre_process).batch(batch_size)

network = Sequential([layers.Dense(256, activation='relu'),
                      layers.Dense(128, activation='relu'),
                      layers.Dense(64, activation='relu'),
                      layers.Dense(32, activation='relu'),
                      layers.Dense(10)])
network.build(input_shape=(None, 28 * 28))
network.summary()

network.compile(optimizer=optimizers.Adam(lr=0.01),
                loss=tf.losses.CategoricalCrossentropy(from_logits=True),
                metrics=['accuracy']
                )

network.fit(db, epochs=3, validation_data=ds_val, validation_freq=2)

network.evaluate(ds_val)
# 保存模型
network.save('model/model.h5')
print('saved total model.')
del network

# 加载模型
print('loaded model from file.')
network = tf.keras.models.load_model('model/model.h5', compile=False)

network.compile(optimizer=optimizers.Adam(lr=0.01),
                loss=tf.losses.CategoricalCrossentropy(from_logits=True),
                metrics=['accuracy']
                )

network.evaluate(ds_val)

4.3. saved_model

5. Keras实战CIFAR10

5.1. CIFAR10

5.2. My Dense layer

5.3. Code

import tensorflow as tf
from tensorflow import keras
from tensorflow.keras import datasets, layers, optimizers, Sequential, metrics
import os

os.environ['TF_CPP_MIN_LOG_LEVEL'] = '3'


def pre_process(x, y):
    x = tf.cast(x, dtype=tf.float32) / 255.
    y = tf.cast(y, dtype=tf.int32)

    return x, y


batch_size = 128
# (50000, 32, 32, 3) (50000, 1)
(x, y), (x_val, y_val) = datasets.cifar10.load_data()
# (50000, 32, 32, 3) (50000,)
y = tf.squeeze(y)
y_val = tf.squeeze(y_val)
# (50000, 32, 32, 3) (50000, 10)
y = tf.one_hot(y, depth=10)
y_val = tf.one_hot(y_val, depth=10)

train_db = tf.data.Dataset.from_tensor_slices((x, y))
train_db = train_db.map(pre_process).shuffle(10000).batch(batch_size)

test_db = tf.data.Dataset.from_tensor_slices((x_val, y_val))
test_db = test_db.map(pre_process).batch(batch_size)

sample = next(iter(train_db))
# batch:  (128, 32, 32, 3) (128, 10)
print("batch: ", sample[0].shape, sample[1].shape)


class MyDense(layers.Layer):
    """自定义层"""

    def __init__(self, input_dim, out_dim):
        super(MyDense, self).__init__()
        self.kernel = self.add_variable('w', [input_dim, out_dim])
        # 去掉偏置
        # self.bias = self.add_variable('b', [out_dim])

    def call(self, inputs, training=None):
        """前向传播的函数"""
        x = inputs @ self.kernel
        return x


class MyNetwork(keras.Model):

    def __init__(self):
        super(MyNetwork, self).__init__()
        self.fc1 = MyDense(32*32*3, 256)
        self.fc2 = MyDense(256, 128)
        self.fc3 = MyDense(128, 64)
        self.fc4 = MyDense(64, 32)
        self.fc5 = MyDense(32, 10)

    def call(self, inputs, training=None):
        """
        前向传播
        inputs: [b, 32, 32, 3]
        """
        x = tf.reshape(inputs, [-1, 32*32*3])

        x = self.fc1(x)
        x = tf.nn.relu(x)

        x = self.fc2(x)
        x = tf.nn.relu(x)

        x = self.fc3(x)
        x = tf.nn.relu(x)

        x = self.fc4(x)
        x = tf.nn.relu(x)

        x = self.fc5(x)

        return x


network = MyNetwork()
network.compile(optimizer=optimizers.Adam(lr=1e-3),
                loss=tf.losses.CategoricalCrossentropy(from_logits=True),
                metrics=['accuracy'])
network.fit(train_db, epochs=5, validation_data=test_db, validation_freq=1)
# 模型的保存与加载测试
network.evaluate(test_db)
network.save_weights('ckpt/weights.ckpt')
del network
print('saved to ckpt/weights.ckpt')

network = MyNetwork()
network.compile(optimizer=optimizers.Adam(lr=1e-3),
                loss=tf.losses.CategoricalCrossentropy(from_logits=True),
                metrics=['accuracy'])
network.load_weights('ckpt/weights.ckpt')
print('loaded weights from file.')
network.evaluate(test_db)

6. 需要全套课程视频+PPT+代码资源可以私聊我

• 方式1：CSDN私信我!
• 方式2：QQ邮箱:[email protected]或者直接加我QQ: 594042358!