Question

我对tensorflow很新，我正在尝试将我的.pb（proto buffer）文件转换为lite版本。但我面临一些问题。导入时间，系统，警告，glob，随机，cv2，base64，json，csv，os

import numpy as np
import tensorflow as tf
from collections import OrderedDict
def load_graph(frozen_graph_filename):
    with tf.gfile.GFile(frozen_graph_filename, "rb") as f:
        graph_def = tf.GraphDef()
        graph_def.ParseFromString(f.read())
    with tf.Graph().as_default() as graph:
        tf.import_graph_def(
            graph_def, 
            input_map=None, 
            return_elements=None, 
            name="prefix", 
            op_dict=None, 
            producer_op_list=None
        )
    return graph

此函数为我加载图形，现在我想将此图形转换为tflite，我使用了以下脚本。

CD_graph = load_graph("CD_Check_k.pb")
CD_input = CD_graph.get_tensor_by_name('prefix/input_node:0')
CD_output = CD_graph.get_tensor_by_name('prefix/output_node:0')
x_single = tf.placeholder(tf.float32, [1, 256 , 256, 3],
                              name="input_node")
with tf.Session() as sess:
  tflite_model = tf.contrib.lite.toco_convert(CD_graph, input_tensors=[x_single ], output_tensors=[CD_output])
  with open('./mnist.tflite', "wb") as f:
      f.write(tflite_model)

错误消息：

'Graph' object has no attribute 'SerializeToString'

Answer 1

您可以使用TocoConverter.from_frozen_graph() API简化代码，因此您不再需要读取冻结的图形。 documentation的示例复制如下。

从文件导出GraphDef

以下示例显示了将GraphDef存储在文件中时如何将TensorFlow GraphDef转换为TensorFlow Lite FlatBuffer。 .pb和.pbtxt文件都被接受。

该示例使用Mobilenet_1.0_224。该功能仅支持通过freeze_graph.py冻结的GraphDef。

import tensorflow as tf

graph_def_file = "/path/to/Downloads/mobilenet_v1_1.0_224/frozen_graph.pb"
input_arrays = ["input"]
output_arrays = ["MobilenetV1/Predictions/Softmax"]

converter = tf.contrib.lite.TocoConverter.from_frozen_graph(
  graph_def_file, input_arrays, output_arrays)
tflite_model = converter.convert()
open("converted_model.tflite", "wb").write(tflite_model)

加载图表，然后用它来构建tflite？

1 个答案:

从文件导出GraphDef