tf2.x 保存模型 加载模型 .h5或.pb转为.tflite

保存模型

以下三种方法均可以保存模型结构和参数
①保存为.h5

net.save('net.h5',save_format='h5')

②保存为.pb

tf.saved_model.save(net,"pb1/")

③保存为.pb

tf.keras.models.save_model(net,'pb2/')

加载模型

   new_model = tf.keras.models.load_model('net.h5')
   new_model = tf.keras.models.load_model("pb1")
   new_model = tf.keras.models.load_model('pb2')

模型转换

① .h5 转 .tflite

loaded_keras_model = tf.keras.models.load_model('net.h5')
keras_to_tflite_converter = 
			tf.lite.TFLiteConverter.from_keras_model(loaded_keras_model)
keras_tflite = keras_to_tflite_converter.convert()

with open('net.tflite', 'wb') as f:
    f.write(keras_tflite)

② .pb转.tflite

saved_model_to_tflite_converter = 
			tf.lite.TFLiteConverter.from_saved_model('pb1')
saved_model_tflite = saved_model_to_tflite_converter.convert()

with open('net.tflite', 'wb') as f:
    f.write(saved_model_tflite)