Example not working (anymore?)

[cg-laser]

I can't get this example to work. I'm not sure if this is because of inconsistencies with newer dependencies but everything I tried was unsuccessful. This is what I tried so far:

The first error I get when following the example step by step is

Traceback (most recent call last):
  File "/Users/cglaser/work/ARIANNA/analysis/2019deeplearning/classification/uTensor_examples/simple_cnn_tutorial/.venv/bin/utensor-cli", line 5, in <module>
    from utensor_cgen.cli import cli
  File "/Users/cglaser/work/ARIANNA/analysis/2019deeplearning/classification/uTensor_examples/simple_cnn_tutorial/.venv/lib/python2.7/site-packages/utensor_cgen/cli.py", line 8, in <module>
    from .utils import NArgsParam
  File "/Users/cglaser/work/ARIANNA/analysis/2019deeplearning/classification/uTensor_examples/simple_cnn_tutorial/.venv/lib/python2.7/site-packages/utensor_cgen/utils.py", line 11, in <module>
    from tensorflow.tools.graph_transforms import TransformGraph
ImportError: No module named graph_transforms

which is because tensorflow 2.1.0 was installed by pip install utensor_cgen==0.3.3.dev2

So I downgraded tensorflow to 1.13.2 but then I just run into the next error:

utensor-cli convert cifar10_cnn.pb --output-nodes=fully_connect_2/logits

WARNING: The TensorFlow contrib module will not be included in TensorFlow 2.0.
For more information, please see:
  * https://github.com/tensorflow/community/blob/master/rfcs/20180907-contrib-sunset.md
  * https://github.com/tensorflow/addons
If you depend on functionality not listed there, please file an issue.

Traceback (most recent call last):
  File "/Users/cglaser/work/ARIANNA/analysis/2019deeplearning/classification/uTensor_examples/simple_cnn_tutorial/.venv/bin/utensor-cli", line 8, in <module>
    sys.exit(cli())
  File "/Users/cglaser/work/ARIANNA/analysis/2019deeplearning/classification/uTensor_examples/simple_cnn_tutorial/.venv/lib/python2.7/site-packages/click/core.py", line 764, in __call__
    return self.main(*args, **kwargs)
  File "/Users/cglaser/work/ARIANNA/analysis/2019deeplearning/classification/uTensor_examples/simple_cnn_tutorial/.venv/lib/python2.7/site-packages/click/core.py", line 717, in main
    rv = self.invoke(ctx)
  File "/Users/cglaser/work/ARIANNA/analysis/2019deeplearning/classification/uTensor_examples/simple_cnn_tutorial/.venv/lib/python2.7/site-packages/click/core.py", line 1137, in invoke
    return _process_result(sub_ctx.command.invoke(sub_ctx))
  File "/Users/cglaser/work/ARIANNA/analysis/2019deeplearning/classification/uTensor_examples/simple_cnn_tutorial/.venv/lib/python2.7/site-packages/click/core.py", line 956, in invoke
    return ctx.invoke(self.callback, **ctx.params)
  File "/Users/cglaser/work/ARIANNA/analysis/2019deeplearning/classification/uTensor_examples/simple_cnn_tutorial/.venv/lib/python2.7/site-packages/click/core.py", line 555, in invoke
    return callback(*args, **kwargs)
  File "/Users/cglaser/work/ARIANNA/analysis/2019deeplearning/classification/uTensor_examples/simple_cnn_tutorial/.venv/lib/python2.7/site-packages/utensor_cgen/cli.py", line 84, in convert_graph
    generator.generate(model_path)
  File "/Users/cglaser/work/ARIANNA/analysis/2019deeplearning/classification/uTensor_examples/simple_cnn_tutorial/.venv/lib/python2.7/site-packages/utensor_cgen/backend/code_generator.py", line 49, in generate
    parser_cls = FrontendSelector.select_parser(ext)
  File "/Users/cglaser/work/ARIANNA/analysis/2019deeplearning/classification/uTensor_examples/simple_cnn_tutorial/.venv/lib/python2.7/site-packages/utensor_cgen/frontend/__init__.py", line 26, in select_parser
    cls._setup()
  File "/Users/cglaser/work/ARIANNA/analysis/2019deeplearning/classification/uTensor_examples/simple_cnn_tutorial/.venv/lib/python2.7/site-packages/utensor_cgen/frontend/__init__.py", line 46, in _setup
    importlib.import_module(mod_name)
  File "/Users/cglaser/.pyenv/versions/2.7.15/lib/python2.7/importlib/__init__.py", line 37, in import_module
    __import__(name)
  File "/Users/cglaser/work/ARIANNA/analysis/2019deeplearning/classification/uTensor_examples/simple_cnn_tutorial/.venv/lib/python2.7/site-packages/utensor_cgen/frontend/onnx.py", line 2, in <module>
    from onnx_tf.backend import prepare
  File "/Users/cglaser/work/ARIANNA/analysis/2019deeplearning/classification/uTensor_examples/simple_cnn_tutorial/.venv/lib/python2.7/site-packages/onnx_tf/__init__.py", line 1, in <module>
    from . import backend
  File "/Users/cglaser/work/ARIANNA/analysis/2019deeplearning/classification/uTensor_examples/simple_cnn_tutorial/.venv/lib/python2.7/site-packages/onnx_tf/backend.py", line 29, in <module>
    from onnx_tf.common.handler_helper import get_all_backend_handlers
  File "/Users/cglaser/work/ARIANNA/analysis/2019deeplearning/classification/uTensor_examples/simple_cnn_tutorial/.venv/lib/python2.7/site-packages/onnx_tf/common/handler_helper.py", line 5, in <module>
    from onnx_tf.handlers.backend import *  # noqa
  File "/Users/cglaser/work/ARIANNA/analysis/2019deeplearning/classification/uTensor_examples/simple_cnn_tutorial/.venv/lib/python2.7/site-packages/onnx_tf/handlers/backend/scatter_nd.py", line 8, in <module>
    @tf_func(tf.tensor_scatter_nd_update)
AttributeError: 'module' object has no attribute 'tensor_scatter_nd_update'
make: *** [models/cifar10_cnn.cpp] Error 1

I get the same error also for python 3.7.3.

I then tried upgrading utensor_cgen to the latest release, but this also crashes maybe because it is not python2.7 compatible anymore?

pip install utensor_cgen --upgrade
DEPRECATION: Python 2.7 will reach the end of its life on January 1st, 2020. Please upgrade your Python as Python 2.7 won't be maintained after that date. A future version of pip will drop support for Python 2.7. More details about Python 2 support in pip, can be found at https://pip.pypa.io/en/latest/development/release-process/#python-2-support
Collecting utensor_cgen
  Using cached https://files.pythonhosted.org/packages/a2/a3/d6db368011e481e38dce7b209d3818942b5dc157ef8b528b8cf3bf414d64/utensor_cgen-0.4.2.tar.gz
    ERROR: Command errored out with exit status 1:
     command: /Users/cglaser/work/ARIANNA/analysis/2019deeplearning/classification/uTensor_examples/simple_cnn_tutorial/.venv/bin/python2.7 -c 'import sys, setuptools, tokenize; sys.argv[0] = '"'"'/private/var/folders/bt/vcs5g33x3mn05hx6tn023ssc0000gn/T/pip-install-OxAhcH/utensor-cgen/setup.py'"'"'; __file__='"'"'/private/var/folders/bt/vcs5g33x3mn05hx6tn023ssc0000gn/T/pip-install-OxAhcH/utensor-cgen/setup.py'"'"';f=getattr(tokenize, '"'"'open'"'"', open)(__file__);code=f.read().replace('"'"'\r\n'"'"', '"'"'\n'"'"');f.close();exec(compile(code, __file__, '"'"'exec'"'"'))' egg_info --egg-base /private/var/folders/bt/vcs5g33x3mn05hx6tn023ssc0000gn/T/pip-install-OxAhcH/utensor-cgen/pip-egg-info
         cwd: /private/var/folders/bt/vcs5g33x3mn05hx6tn023ssc0000gn/T/pip-install-OxAhcH/utensor-cgen/
    Complete output (24 lines):
    DEPRECATION: Python 2.7 will reach the end of its life on January 1st, 2020. Please upgrade your Python as Python 2.7 won't be maintained after that date. A future version of pip will drop support for Python 2.7. More details about Python 2 support in pip, can be found at https://pip.pypa.io/en/latest/development/release-process/#python-2-support
    Traceback (most recent call last):
      File "<string>", line 1, in <module>
      File "/private/var/folders/bt/vcs5g33x3mn05hx6tn023ssc0000gn/T/pip-install-OxAhcH/utensor-cgen/setup.py", line 79, in <module>
        "Topic :: Utilities"
      File "/Users/cglaser/work/ARIANNA/analysis/2019deeplearning/classification/uTensor_examples/simple_cnn_tutorial/.venv/lib/python2.7/site-packages/setuptools/__init__.py", line 145, in setup
        return distutils.core.setup(**attrs)
      File "/Users/cglaser/.pyenv/versions/2.7.15/lib/python2.7/distutils/core.py", line 111, in setup
        _setup_distribution = dist = klass(attrs)
      File "/Users/cglaser/work/ARIANNA/analysis/2019deeplearning/classification/uTensor_examples/simple_cnn_tutorial/.venv/lib/python2.7/site-packages/setuptools/dist.py", line 448, in __init__
        k: v for k, v in attrs.items()
      File "/Users/cglaser/.pyenv/versions/2.7.15/lib/python2.7/distutils/dist.py", line 287, in __init__
        self.finalize_options()
      File "/Users/cglaser/work/ARIANNA/analysis/2019deeplearning/classification/uTensor_examples/simple_cnn_tutorial/.venv/lib/python2.7/site-packages/setuptools/dist.py", line 740, in finalize_options
        ep.load()(self)
      File "/Users/cglaser/work/ARIANNA/analysis/2019deeplearning/classification/uTensor_examples/simple_cnn_tutorial/.venv/lib/python2.7/site-packages/setuptools/dist.py", line 747, in _finalize_setup_keywords
        ep.load()(self, ep.name, value)
      File "/private/var/folders/bt/vcs5g33x3mn05hx6tn023ssc0000gn/T/pip-install-OxAhcH/utensor-cgen/.eggs/better_setuptools_git_version-1.0.5-py2.7.egg/better_setuptools_git_version.py", line 82, in validate_version_config
        dist.metadata.version = get_version(template, starting_version)
      File "/private/var/folders/bt/vcs5g33x3mn05hx6tn023ssc0000gn/T/pip-install-OxAhcH/utensor-cgen/.eggs/better_setuptools_git_version-1.0.5-py2.7.egg/better_setuptools_git_version.py", line 52, in get_version
        tag = get_tag()
      File "/private/var/folders/bt/vcs5g33x3mn05hx6tn023ssc0000gn/T/pip-install-OxAhcH/utensor-cgen/.eggs/better_setuptools_git_version-1.0.5-py2.7.egg/better_setuptools_git_version.py", line 9, in get_tag
        return subprocess.getoutput("git tag --sort=version:refname --merged | tail -n1")
    AttributeError: 'module' object has no attribute 'getoutput'
    ----------------------------------------
ERROR: Command errored out with exit status 1: python setup.py egg_info Check the logs for full command output.

With python3 I upgraded utensor_cgen to the latest version which also downgraded tensorflow to 1.13.1
Successfully installed graphviz-0.13.2 onnx-tf-1.2.1 tensorflow-1.13.1 toml-0.10.0 utensor-cgen-0.4.2

This allows to execute utensor-cli convert cifar10_cnn.pb --output-nodes fully_connect_2/logits without errors.

However, the compilation of the mbed code then crashes with

mbed compile -m auto -t GCC_ARM --profile=uTensor/build_profile/release.json -f
[mbed] Working path "/Users/cglaser/work/ARIANNA/analysis/2019deeplearning/classification/uTensor_examples/simple_cnn_tutorial_p3" (program)
[mbed] WARNING: Could not find mbed program in current path "/Users/cglaser/work/ARIANNA/analysis/2019deeplearning/classification/uTensor_examples/simple_cnn_tutorial_p3".
       You can fix this by calling "mbed new ." in the root of your program.
---
[mbed] Detected "K64F" connected to "/Volumes/DAPLINK" and using com port "/dev/tty.usbmodem1422"
[Warning] @,: Compiler version mismatch: Have 9.2.1; expected version >= 6.0.0 and < 7.0.0
Building project simple_cnn_tutorial_p3 (K64F, GCC_ARM)
Scan: simple_cnn_tutorial_p3
...
Compile [ 85.2%]: mbed_rtc_time.cpp
[Error] mbed_rtc_time.cpp@89,22: invalid use of incomplete type 'const struct timeval'
[Error] mbed_rtc_time.cpp@110,7: invalid use of incomplete type 'struct timeval'
[Error] mbed_rtc_time.cpp@111,7: invalid use of incomplete type 'struct timeval'
[Error] mbed_rtc_time.cpp@124,20: aggregate 'timeval tv' has incomplete type and cannot be defined
[Error] mbed_rtc_time.cpp@137,26: variable 'const timeval tv' has initializer but incomplete type
[ERROR] '_queue.SimpleQueue' object has no attribute 'queue'
[mbed] ERROR: "/Users/cglaser/work/ARIANNA/analysis/2019deeplearning/classification/uTensor_examples/simple_cnn_tutorial_p3/.venv/bin/python" returned error.
       Code: 1
       Path: "/Users/cglaser/work/ARIANNA/analysis/2019deeplearning/classification/uTensor_examples/simple_cnn_tutorial_p3"
       Command: "/Users/cglaser/work/ARIANNA/analysis/2019deeplearning/classification/uTensor_examples/simple_cnn_tutorial_p3/.venv/bin/python -u /Users/cglaser/work/ARIANNA/analysis/2019deeplearning/classification/uTensor_examples/simple_cnn_tutorial_p3/mbed-os/tools/make.py -t GCC_ARM -m K64F --profile uTensor/build_profile/release.json --source . --build ./BUILD/K64F/GCC_ARM-RELEASE"
       Tip: You could retry the last command with "-v" flag for verbose output
---
make: *** [compile] Error 255

Maybe this is because my compiler version is too high? [Warning] @,: Compiler version mismatch: Have 9.2.1; expected version >= 6.0.0 and < 7.0.0
So I downgraded my GCC_ARM toolchain to 6.3.1 but I still get the same error.

Any help would be greatly appreciated!

[finnss]

I'm having a similar issue and would love to see an answer!

As a side note, how did you downgrade your toolchain? I've downloaded GCC ARM None Eabi v9.2.1, and I tried pointing to it in several ways.

• mbed config -G GCC_ARM_PATH ~/Documents/skole/master_thesis/arm_mbed/gcc-arm
• The same without -G.
• The same but using MBED_GCC_ARM_PATH, I'm still not sure which is used
• The same, but pointing to the /bin folder of that path
• Removing all mbed config and choosing the parameters through environment variables

None of these attempts seemed to change anything, though. I'm getting [Warning] @,: Compiler version mismatch: Have 6.3.1; expected version >= 9.0.0 and < 10.0.0 no matter what I do. Also, isn't it weird that it is requiring a version between 9 and 10? Most threads I've seen say only version 6 is supported.

[finnss]

For the record, I'm working on a different uTensor example, the one from this guide: https://www.hackster.io/news/simple-neural-network-on-mcus-a7cbd3dc108c

I'm guessing the underlying issue is the same though, and I didn't have any better luck trying this simple_cnn_tutorial. My error message is slightly different:

mbed compile --target NRF52_DK --toolchain GCC_ARM -f                                                                 finn@Finns-MacBook-Pro
[mbed] Working path "/Users/finn/Documents/skole/master_thesis/arm_mbed/my_uTensor" (library)
[mbed] Program path "/Users/finn/Documents/skole/master_thesis/arm_mbed/my_uTensor"
[Warning] @,: Compiler version mismatch: Have 6.3.1; expected version >= 9.0.0 and < 10.0.0
Building project my_uTensor (NRF52_DK, GCC_ARM)
Scan: my_uTensor
Compile [  0.3%]: ESP8266.cpp
[Error] ctime@62,11: '::tm' has not been declared
[Error] ctime@64,11: '::clock' has not been declared
[Error] ctime@65,11: '::difftime' has not been declared
[Error] ctime@66,11: '::mktime' has not been declared
[Error] ctime@67,11: '::time' has not been declared
[Error] ctime@68,11: '::asctime' has not been declared
[Error] ctime@69,11: '::ctime' has not been declared
[Error] ctime@70,11: '::gmtime' has not been declared
[Error] ctime@71,11: '::localtime' has not been declared
[Error] ctime@72,11: '::strftime' has not been declared
[Error] time.h@29,32: 'tm' was not declared in this scope
[Error] time.h@29,36: expected primary-expression before 'const'
[Error] time.h@34,33: variable or field 'LocalTime' declared void
[Error] time.h@34,33: 'tm' was not declared in this scope
[Error] time.h@34,37: 't' was not declared in this scope
[Error] time.h@34,52: expected primary-expression before 'time'
[Error] time.h@39,30: variable or field 'GMTime' declared void
[Error] time.h@39,30: 'tm' was not declared in this scope
[Error] time.h@39,34: 't' was not declared in this scope
[Error] time.h@39,49: expected primary-expression before 'time'
[ERROR] In file included from ./venv/lib/python3.7/site-packages/tensorflow/include/external/aws/aws-cpp-sdk-core/include/aws/core/platform/time.h:19:0,
                 from ./mbed-os/platform/mbed_retarget.h:80,
                 from ./mbed-os/platform/platform.h:26,
                 from ./mbed-os/drivers/UARTSerial.h:21,
                 from ./mbed-os/components/wifi/esp8266-driver/ESP8266/ESP8266.h:23,
                 from ./mbed-os/components/wifi/esp8266-driver/ESP8266/ESP8266.cpp:26:
/usr/local/Cellar/arm-gcc-bin/6-2017-q2-update/arm-none-eabi/include/c++/6.3.1/ctime:62:11: error: '::tm' has not been declared
   using ::tm;
           ^~
/usr/local/Cellar/arm-gcc-bin/6-2017-q2-update/arm-none-eabi/include/c++/6.3.1/ctime:64:11: error: '::clock' has not been declared
   using ::clock;
           ^~~~~
/usr/local/Cellar/arm-gcc-bin/6-2017-q2-update/arm-none-eabi/include/c++/6.3.1/ctime:65:11: error: '::difftime' has not been declared
   using ::difftime;
           ^~~~~~~~
/usr/local/Cellar/arm-gcc-bin/6-2017-q2-update/arm-none-eabi/include/c++/6.3.1/ctime:66:11: error: '::mktime' has not been declared
   using ::mktime;
           ^~~~~~
/usr/local/Cellar/arm-gcc-bin/6-2017-q2-update/arm-none-eabi/include/c++/6.3.1/ctime:67:11: error: '::time' has not been declared
   using ::time;
           ^~~~
/usr/local/Cellar/arm-gcc-bin/6-2017-q2-update/arm-none-eabi/include/c++/6.3.1/ctime:68:11: error: '::asctime' has not been declared
   using ::asctime;
           ^~~~~~~
/usr/local/Cellar/arm-gcc-bin/6-2017-q2-update/arm-none-eabi/include/c++/6.3.1/ctime:69:11: error: '::ctime' has not been declared
   using ::ctime;
           ^~~~~
/usr/local/Cellar/arm-gcc-bin/6-2017-q2-update/arm-none-eabi/include/c++/6.3.1/ctime:70:11: error: '::gmtime' has not been declared
   using ::gmtime;
           ^~~~~~
/usr/local/Cellar/arm-gcc-bin/6-2017-q2-update/arm-none-eabi/include/c++/6.3.1/ctime:71:11: error: '::localtime' has not been declared
   using ::localtime;
           ^~~~~~~~~
/usr/local/Cellar/arm-gcc-bin/6-2017-q2-update/arm-none-eabi/include/c++/6.3.1/ctime:72:11: error: '::strftime' has not been declared
   using ::strftime;
           ^~~~~~~~
In file included from ./mbed-os/platform/mbed_retarget.h:80:0,
                 from ./mbed-os/platform/platform.h:26,
                 from ./mbed-os/drivers/UARTSerial.h:21,
                 from ./mbed-os/components/wifi/esp8266-driver/ESP8266/ESP8266.h:23,
                 from ./mbed-os/components/wifi/esp8266-driver/ESP8266/ESP8266.cpp:26:
./venv/lib/python3.7/site-packages/tensorflow/include/external/aws/aws-cpp-sdk-core/include/aws/core/platform/time.h:29:32: error: 'tm' was not declared in this scope
     AWS_CORE_API time_t TimeGM(tm* const t);
                                ^~
./venv/lib/python3.7/site-packages/tensorflow/include/external/aws/aws-cpp-sdk-core/include/aws/core/platform/time.h:29:36: error: expected primary-expression before 'const'
     AWS_CORE_API time_t TimeGM(tm* const t);
                                    ^~~~~
./venv/lib/python3.7/site-packages/tensorflow/include/external/aws/aws-cpp-sdk-core/include/aws/core/platform/time.h:34:33: error: variable or field 'LocalTime' declared void
     AWS_CORE_API void LocalTime(tm* t, std::time_t time);
                                 ^~
./venv/lib/python3.7/site-packages/tensorflow/include/external/aws/aws-cpp-sdk-core/include/aws/core/platform/time.h:34:33: error: 'tm' was not declared in this scope
./venv/lib/python3.7/site-packages/tensorflow/include/external/aws/aws-cpp-sdk-core/include/aws/core/platform/time.h:34:37: error: 't' was not declared in this scope
     AWS_CORE_API void LocalTime(tm* t, std::time_t time);
                                     ^
./venv/lib/python3.7/site-packages/tensorflow/include/external/aws/aws-cpp-sdk-core/include/aws/core/platform/time.h:34:52: error: expected primary-expression before 'time'
     AWS_CORE_API void LocalTime(tm* t, std::time_t time);
                                                    ^~~~
./venv/lib/python3.7/site-packages/tensorflow/include/external/aws/aws-cpp-sdk-core/include/aws/core/platform/time.h:39:30: error: variable or field 'GMTime' declared void
     AWS_CORE_API void GMTime(tm* t, std::time_t time);
                              ^~
./venv/lib/python3.7/site-packages/tensorflow/include/external/aws/aws-cpp-sdk-core/include/aws/core/platform/time.h:39:30: error: 'tm' was not declared in this scope
./venv/lib/python3.7/site-packages/tensorflow/include/external/aws/aws-cpp-sdk-core/include/aws/core/platform/time.h:39:34: error: 't' was not declared in this scope
     AWS_CORE_API void GMTime(tm* t, std::time_t time);
                                  ^
./venv/lib/python3.7/site-packages/tensorflow/include/external/aws/aws-cpp-sdk-core/include/aws/core/platform/time.h:39:49: error: expected primary-expression before 'time'
     AWS_CORE_API void GMTime(tm* t, std::time_t time);
                                                 ^~~~

[mbed] ERROR: "/Users/finn/Documents/skole/master_thesis/arm_mbed/my_uTensor/venv/bin/python3" returned error.
       Code: 1
       Path: "/Users/finn/Documents/skole/master_thesis/arm_mbed/my_uTensor"
       Command: "/Users/finn/Documents/skole/master_thesis/arm_mbed/my_uTensor/venv/bin/python3 -u /Users/finn/Documents/skole/master_thesis/arm_mbed/my_uTensor/mbed-os/tools/make.py -t GCC_ARM -m NRF52_DK --source . --build ./BUILD/NRF52_DK/GCC_ARM"
       Tip: You could retry the last command with "-v" flag for verbose output