python - Tensorflow [Toco] 将模型转换为优化格式导致 ValueError

Question

我刚开始尝试Tensorflow，我想在Android上做一些实验，但我读到Tensorflow for Poets（移动版）支持.tflite，而我下载并发现谷歌搜索的模型是.pb。

所以我决定使用toco转换它们，但是当我执行转换命令时，我总是得到这个错误

ValueError：找到了无效的张量“输入”。

CLI 命令是：

toco   --graph_def_file=tensorflow_inception_graph.pb   --output_file=optimized_graph.tflite   --input_format=TENSORFLOW_GRAPHDEF --output_format=TFLITE   --input_shape=1,300,300,3   --input_array=input   --o
utput_array=final_result   --inference_type=FLOAT   --inference_input_type=FLOAT

那有什么问题？

堆栈跟踪

File "/home/Mitro/venv/bin/toco", line 11, in <module>
    sys.exit(main())
  File "/home/Mitro/venv/lib/python3.5/site-packages/tensorflow/contrib/lite/python/tflite_convert.py", line 401, in main
    app.run(main=run_main, argv=sys.argv[:1])
  File "/home/Mitro/venv/lib/python3.5/site-packages/tensorflow/python/platform/app.py", line 125, in run
    _sys.exit(main(argv))
  File "/home/Mitro/venv/lib/python3.5/site-packages/tensorflow/contrib/lite/python/tflite_convert.py", line 397, in run_main
    _convert_model(tflite_flags)
  File "/home/Mitro/venv/lib/python3.5/site-packages/tensorflow/contrib/lite/python/tflite_convert.py", line 100, in _convert_model
    converter = _get_toco_converter(flags)
  File "/home/Mitro/venv/lib/python3.5/site-packages/tensorflow/contrib/lite/python/tflite_convert.py", line 87, in _get_toco_converter
    return converter_fn(**converter_kwargs)
  File "/home/Mitro/venv/lib/python3.5/site-packages/tensorflow/contrib/lite/python/lite.py", line 271, in from_frozen_graph
    sess.graph, input_arrays)
  File "/home/Mitro/venv/lib/python3.5/site-packages/tensorflow/contrib/lite/python/convert_saved_model.py", line 189, in get_tensors_from_tensor_names
    ",".join(invalid_tensors)))

我来自谷歌的参考

标签： pythontensorflowtensorflow-litetoco

Answer 1

运行此脚本，将 model_file 参数更改为 pb 文件，作为输出，您将获得如下内容：

Tensor("import/ Placeholder :0", shape=(?, 299, 299, 3), dtype=float32)
Tensor("import/ final_result :0", shape=(?, 5), dtype=float32)

在这种情况下，它们是您必须通过的参数
--input_array = Placeholder
--output_array = final_result

这是完整的信息