加载 keras model h5 未知指标

Question

我已经训练了一个 keras CNN 监控如下指标：

METRICS = [
  TruePositives(name='tp'),
  FalsePositives(name='fp'),
  TrueNegatives(name='tn'),
  FalseNegatives(name='fn'), 
  BinaryAccuracy(name='accuracy'),
  Precision(name='precision'),
  Recall(name='recall'),
  AUC(name='auc'),
 ]

然后是 model.compile：

 model.compile(optimizer='nadam', loss='binary_crossentropy',
         metrics=METRICS)

它工作得很好，我保存了我的 h5 model (model.h5)。

现在我已经下载了 model 并且我想在导入 model 的其他脚本中使用它：

 from keras.models import load_model
 model = load_model('model.h5')
 model.predict(....)

但在运行期间编译器返回：

 ValueError: Unknown metric function: {'class_name': 'TruePositives', 'config': {'name': 'tp', 'dtype': 'float32', 'thresholds': None}}

我应该如何处理这个问题？

先感谢您

Answer 1

当您有自定义指标时，您需要遵循稍微不同的方法。

1. 创建 model，训练并保存 model
2. 使用 custom_objects 加载custom_objects并compile = False
3. 最后用 custom_objects 编译 model

我在这里展示了方法

import tensorflow as tf
from tensorflow import keras
mnist = tf.keras.datasets.mnist

(x_train, y_train),(x_test, y_test) = mnist.load_data()
x_train, x_test = x_train / 255.0, x_test / 255.0

# Custom Loss1 (for example) 
#@tf.function() 
def customLoss1(yTrue,yPred):
  return tf.reduce_mean(yTrue-yPred) 

# Custom Loss2 (for example) 
#@tf.function() 
def customLoss2(yTrue, yPred):
  return tf.reduce_mean(tf.square(tf.subtract(yTrue,yPred))) 

def create_model():
  model = tf.keras.models.Sequential([
    tf.keras.layers.Flatten(input_shape=(28, 28)),
    tf.keras.layers.Dense(512, activation=tf.nn.relu),  
    tf.keras.layers.Dropout(0.2),
    tf.keras.layers.Dense(10, activation=tf.nn.softmax)
    ])
  model.compile(optimizer='adam',
              loss='sparse_categorical_crossentropy',
              metrics=['accuracy', customLoss1, customLoss2])
  return model 

# Create a basic model instance
model=create_model()

# Fit and evaluate model 
model.fit(x_train, y_train, epochs=5)

loss, acc,loss1, loss2 = model.evaluate(x_test, y_test,verbose=1)
print("Original model, accuracy: {:5.2f}%".format(100*acc)) # Original model, accuracy: 98.11%

# saving the model
model.save('./Mymodel',save_format='tf')

# load the model
loaded_model = tf.keras.models.load_model('./Mymodel',custom_objects={'customLoss1':customLoss1,'customLoss2':customLoss2},compile=False)

# compile the model
loaded_model.compile(optimizer='adam',
              loss='sparse_categorical_crossentropy',
              metrics=['accuracy', customLoss1, customLoss2])

# loaded model also has same accuracy, metrics and loss
loss, acc,loss1, loss2 = loaded_model.evaluate(x_test, y_test,verbose=1)
print("Loaded model, accuracy: {:5.2f}%".format(100*acc)) #Loaded model, accuracy: 98.11%

Answer 2

看起来您正在玩 tensorflow 教程。 我也使用了这些确切的指标并且遇到了同样的问题。 对我有用的是使用compile = False加载 model，然后使用自定义指标对其进行编译。 然后您应该能够按预期使用model.predict(....) 。

import keras

model = keras.models.load_model('model.h5', compile = False)

METRICS = [
      keras.metrics.TruePositives(name='tp'),
      keras.metrics.FalsePositives(name='fp'),
      keras.metrics.TrueNegatives(name='tn'),
      keras.metrics.FalseNegatives(name='fn'), 
      keras.metrics.BinaryAccuracy(name='accuracy'),
      keras.metrics.Precision(name='precision'),
      keras.metrics.Recall(name='recall'),
      keras.metrics.AUC(name='auc'),
]

model.compile(optimizer = keras.optimizers.Adam(learning_rate=1e-4),
              loss = 'binary_crossentropy',
              metrics = METRICS
             )

Answer 3

custom_objects['METRICS'] = METRICS
model = load_model('model.h5', custom_objects=custom_objects)