如何改变解释器的输出形状？

Question

我想在android上建立一个tflite文件，所以我用jupyter notebook创建了以下模型。

于是我使用jupyter notebook创建了以下模型。

将创建的模型转换为tflite文件后，我们检查它是否被正确转换。

我希望输出的解释器形状结果是 1 这一点，但结果仍然是 110]我应该怎么做？

我做的模型层是这样的。

model = tf.keras.models.Sequential([
  tf.keras.layers.Conv2D(32, (3,3), padding="same", input_shape=X_train.shape[1:], activation="relu"),
  tf.keras.layers.MaxPooling2D(pool_size=(2,2)),
  tf.keras.layers.Conv2D(32, (3,3), padding="same", activation="relu"),
  tf.keras.layers.MaxPooling2D(pool_size=(2,2)),

  tf.keras.layers.Conv2D(64, (3,3), padding="same", activation="relu"),
  tf.keras.layers.MaxPooling2D(pool_size=(2,2)),
  tf.keras.layers.Dropout(0.25),

  tf.keras.layers.Conv2D(64, (3,3), padding="same", activation="relu"),
  tf.keras.layers.MaxPooling2D(pool_size=(2,2)),
  tf.keras.layers.Dropout(0.25),

  tf.keras.layers.Flatten(),
  tf.keras.layers.Dense(256, activation="relu"),
  tf.keras.layers.Dropout(0.5),
  tf.keras.layers.Dense(1, activation="sigmoid")
])

将训练好的模型转换为tflite文件的部分内容

model = tf.keras.models.load_model("./model/model.h5")
converter = tf.lite.TFLiteConverter.from_keras_model(model)
tflite_model = converter.convert()
open("converted_model.tflite", "wb").write(tflite_model)


interpreter = tf.lite.Interpreter(model_path="converted_model.tflite")
interpreter.allocate_tensors()

# Get input and output tensors.
input_details = interpreter.get_input_details()
print("--------------")
print("shape:", input_details[0]['shape'])
print("type:", input_details[0]['dtype'])
output_details = interpreter.get_output_details()
print("--------------")
print("shape:", output_details[0]['shape'])
print("type:", output_details[0]['dtype'])

interpreter.resize_tensor_input(input_details[0]['index'], (39, 64, 64))
interpreter.resize_tensor_input(output_details[0]['index'], (39, 5))
interpreter.allocate_tensors()

input_details = interpreter.get_input_details()
print("--------------")
print("shape:", input_details[0]['shape'])
print("type:", input_details[0]['dtype'])
output_details = interpreter.get_output_details()
print("--------------")
print("shape:", output_details[0]['shape'])
print("type:", output_details[0]['dtype'])

请在此输入图片描述

Answer 1

当用Keras创建模型时，你可以使用。

from tensorflow.keras.layers import Input
from tensorflow.keras.models import Model

# some stuff your code does

input = Input((THE_HEIGHT_YOU_WANT, THE_WIDTH_YOU_WANT, THE_CHANNELS_YOU_WANT))

# all the Tensorflow ops you want on "input"

model = Model(input, THE_OUTPUT_YOU_WANT)

# any other stuff your code might do before saving

model.save(THE_PATH_YOU_WANT)

loaded_model = tf.keras.models.load_model(THE_PATH_YOU_WANT)

# rest of your code for converting and saving the model

现在，当你在Android上运行时，你可以使用：

tensorflowLiteInterpreterInstance.getInputTensor(inputTensorIndex).shape()

来获取模型的形状。

形状应该与 (THE_HEIGHT_YOU_WANT, THE_WIDTH_YOU_WANT, THE_CHANNELS_YOU_WANT) 形状。