windows Docker tengine模型转换工具的安装与使用

windows Docker tengine模型转换工具的安装与使用

最近在开发板上用Tengine做模型推理，遇到了模型不能使用的问题，对模型转换工具安装与使用做一个记录

模型转换工具安装

官网
找一个目录把官网代码clone下来
然后再这个目录

mkdir build && cd build
cmake ..
make -j`nproc` && make install

cmake阶段遇到了如下问题,反正也不用oneflow，就直接加参数了cmake -DBUILD_ONEFLOW_SERIALIZER=OFF ..

root@531776024d67:/home/tengine/build# cmake ..
-- CMAKE_INSTALL_PREFIX = /home/tengine/build/install
-- The CXX compiler identification is GNU 7.5.0
-- Check for working CXX compiler: /usr/bin/c++
-- Check for working CXX compiler: /usr/bin/c++ -- works
-- Detecting CXX compiler ABI info
-- Detecting CXX compiler ABI info - done
-- Detecting CXX compile features
-- Detecting CXX compile features - done
CMake Error at CMakeLists.txt:59 (message):
  Please upgrade your cmake (maybe by "pip3 install -U cmake") or disable
  OneFlow serializer by "cmake -DBUILD_ONEFLOW_SERIALIZER=OFF .."


-- Configuring incomplete, errors occurred!
See also "/home/tengine/build/CMakeFiles/CMakeOutput.log".
See also "/home/tengine/build/CMakeFiles/CMakeError.log".

模型转换工具使用

我的需求是pd转tmfile，直接用模型转换工具未能成功，之后采用pd转tflite,tflite转tmfile

pd转tflite

参考

import tensorflow as tf

# Convert the model
converter = tf.lite.TFLiteConverter.from_saved_model(saved_model_dir) # path to the SavedModel directory
tflite_model = converter.convert()

# Save the model.
with open('model.tflite', 'wb') as f:
   f.write(tflite_model)

tflite转tmfile

 ./install/bin/convert_tool -f tflite -m ../model.tflite -o model.tmfile