Mobilenet_v2

[gordinmitya]

Describe the bug

Unable to convert model from torchvision to pb/tflite. Conversion completes but output file can't be open in neutron app neither executed with tensowflow. If I continue conversion to tflite, it finishes without errors but still file is broken.

To Reproduce

Regular MobilenetV2 model obtained from torchvision as described in official doc link to onnx is below.

given python 3.7 pip install tensorflow-cpu tensorflow-addons onnx install onnx-tensorflow from git as described in readme

you can find Dockerfile and image here

ONNX model file

https://drive.google.com/open?id=1OtQfwWiS8BZ2qW3dOoSbHaaQrZbLMeHL

Python, ONNX, ONNX-TF, Tensorflow version

This section can be obtained by running get_version.py from util folder.

• Python version: 3.7
• ONNX version: 1.6.0
• ONNX-TF version: 1.5.0
• Tensorflow version: 2.1.0

Additional context

onnx-tf convert -i mobilenet_v2.onnx -o model.pb
2020-02-06 13:51:15.663818: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'libnvinfer.so.6'; dlerror: libnvinfer.so.6: cannot open shared object file: No such file or directory
2020-02-06 13:51:15.665546: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'libnvinfer_plugin.so.6'; dlerror: libnvinfer_plugin.so.6: cannot open shared object file: No such file or directory
2020-02-06 13:51:15.666265: W tensorflow/compiler/tf2tensorrt/utils/py_utils.cc:30] Cannot dlopen some TensorRT libraries. If you would like to use Nvidia GPU with TensorRT, please make sure the missing libraries mentioned above are installed properly.
/onnx-tensorflow/onnx_tf/common/__init__.py:89: UserWarning: onnx_tf.common.get_outputs_names is deprecated. It will be removed in future release. Use TensorflowGraph.get_outputs_names instead.
  warnings.warn(message)
2020-02-06 13:51:16.624778: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'libcudart.so.10.1'; dlerror: libcudart.so.10.1: cannot open shared object file: No such file or directory
2020-02-06 13:51:16.626049: I tensorflow/stream_executor/cuda/cudart_stub.cc:29] Ignore above cudart dlerror if you do not have a GPU set up on your machine.
/onnx-tensorflow/onnx_tf/common/handler_helper.py:37: UserWarning: Unknown op ConstantFill in domain `ai.onnx`.
  handler.ONNX_OP, handler.DOMAIN or "ai.onnx"))
/onnx-tensorflow/onnx_tf/common/handler_helper.py:34: UserWarning: Fail to get since_version of ConvInteger in domain `` with max_inclusive_version=9. Set to 1.
  handler.ONNX_OP, handler.DOMAIN, version))
/onnx-tensorflow/onnx_tf/common/handler_helper.py:34: UserWarning: Fail to get since_version of DequantizeLinear in domain `` with max_inclusive_version=9. Set to 1.
  handler.ONNX_OP, handler.DOMAIN, version))
/onnx-tensorflow/onnx_tf/common/handler_helper.py:34: UserWarning: Fail to get since_version of GatherND in domain `` with max_inclusive_version=9. Set to 1.
  handler.ONNX_OP, handler.DOMAIN, version))
/onnx-tensorflow/onnx_tf/common/handler_helper.py:37: UserWarning: Unknown op ImageScaler in domain `ai.onnx`.
  handler.ONNX_OP, handler.DOMAIN or "ai.onnx"))
/onnx-tensorflow/onnx_tf/common/handler_helper.py:34: UserWarning: Fail to get since_version of IsInf in domain `` with max_inclusive_version=9. Set to 1.
  handler.ONNX_OP, handler.DOMAIN, version))
/onnx-tensorflow/onnx_tf/common/handler_helper.py:34: UserWarning: Fail to get since_version of MatMulInteger in domain `` with max_inclusive_version=9. Set to 1.
  handler.ONNX_OP, handler.DOMAIN, version))
/onnx-tensorflow/onnx_tf/common/handler_helper.py:34: UserWarning: Fail to get since_version of Mod in domain `` with max_inclusive_version=9. Set to 1.
  handler.ONNX_OP, handler.DOMAIN, version))
/onnx-tensorflow/onnx_tf/common/handler_helper.py:34: UserWarning: Fail to get since_version of NonMaxSuppression in domain `` with max_inclusive_version=9. Set to 1.
  handler.ONNX_OP, handler.DOMAIN, version))
/onnx-tensorflow/onnx_tf/common/handler_helper.py:34: UserWarning: Fail to get since_version of QLinearConv in domain `` with max_inclusive_version=9. Set to 1.
  handler.ONNX_OP, handler.DOMAIN, version))
/onnx-tensorflow/onnx_tf/common/handler_helper.py:34: UserWarning: Fail to get since_version of QLinearMatMul in domain `` with max_inclusive_version=9. Set to 1.
  handler.ONNX_OP, handler.DOMAIN, version))
/onnx-tensorflow/onnx_tf/common/handler_helper.py:34: UserWarning: Fail to get since_version of QuantizeLinear in domain `` with max_inclusive_version=9. Set to 1.
  handler.ONNX_OP, handler.DOMAIN, version))
/onnx-tensorflow/onnx_tf/common/handler_helper.py:34: UserWarning: Fail to get since_version of Range in domain `` with max_inclusive_version=9. Set to 1.
  handler.ONNX_OP, handler.DOMAIN, version))
/onnx-tensorflow/onnx_tf/common/handler_helper.py:34: UserWarning: Fail to get since_version of Resize in domain `` with max_inclusive_version=9. Set to 1.
  handler.ONNX_OP, handler.DOMAIN, version))
/onnx-tensorflow/onnx_tf/common/handler_helper.py:34: UserWarning: Fail to get since_version of ReverseSequence in domain `` with max_inclusive_version=9. Set to 1.
  handler.ONNX_OP, handler.DOMAIN, version))
/onnx-tensorflow/onnx_tf/common/handler_helper.py:34: UserWarning: Fail to get since_version of Round in domain `` with max_inclusive_version=9. Set to 1.
  handler.ONNX_OP, handler.DOMAIN, version))
/onnx-tensorflow/onnx_tf/common/handler_helper.py:34: UserWarning: Fail to get since_version of ScatterElements in domain `` with max_inclusive_version=9. Set to 1.
  handler.ONNX_OP, handler.DOMAIN, version))
/onnx-tensorflow/onnx_tf/common/handler_helper.py:34: UserWarning: Fail to get since_version of ScatterND in domain `` with max_inclusive_version=9. Set to 1.
  handler.ONNX_OP, handler.DOMAIN, version))
/onnx-tensorflow/onnx_tf/common/handler_helper.py:34: UserWarning: Fail to get since_version of ThresholdedRelu in domain `` with max_inclusive_version=9. Set to 1.
  handler.ONNX_OP, handler.DOMAIN, version))
2020-02-06 13:51:22.098072: I tensorflow/core/platform/cpu_feature_guard.cc:142] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 AVX512F FMA
2020-02-06 13:51:22.117612: I tensorflow/core/platform/profile_utils/cpu_utils.cc:94] CPU Frequency: 2100000000 Hz
2020-02-06 13:51:22.119863: I tensorflow/compiler/xla/service/service.cc:168] XLA service 0x555bc8d37070 initialized for platform Host (this does not guarantee that XLA will be used). Devices:
2020-02-06 13:51:22.121269: I tensorflow/compiler/xla/service/service.cc:176]   StreamExecutor device (0): Host, Default Version
2020-02-06 13:51:22.126981: W tensorflow/stream_executor/platform/default/dso_loader.cc:55] Could not load dynamic library 'libcuda.so.1'; dlerror: libcuda.so.1: cannot open shared object file: No such file or directory
2020-02-06 13:51:22.128099: E tensorflow/stream_executor/cuda/cuda_driver.cc:351] failed call to cuInit: UNKNOWN ERROR (303)
2020-02-06 13:51:22.128811: I tensorflow/stream_executor/cuda/cuda_diagnostics.cc:156] kernel driver does not appear to be running on this host (595ddf8391b4): /proc/driver/nvidia/version does not exist

[vade]

have you tried with tf=1.15 ? I believe thats the last fully supported according to the docs?

Because users often have their own preferences for which variant of Tensorflow to install (i.e., a GPU version instead of a CPU version), we do not explicitly require tensorflow in the installation script. It is therefore users' responsibility to ensure that the proper variant of Tensorflow is available to ONNX-TF. Moreoever, we require Tensorflow version == 1.15.0.

From Readme.md.

Also check out :

https://github.com/onnx/onnx-tensorflow/blob/master/Versioning.md

No idea if that holds up though.

[vade]

Also, im about to try this exact conversion today, so ill let you know how it goes with TF 1.15

[gordinmitya]

@vade Hey, thank you for the response!

I guess the doc is inconsistent a little bit. In installation section: "Install Tensorflow >= 2.0 and tensorflow-addons. (Note for Tensorflow 1.x please refer the tf-1.x branch)". Also I sow in commits "Support Tensorflow 2.0".

Don't remember exact error, but I've tried with 1.15 and it doesn't worked. So please text if you'll find the solution.

[mgarbade]

Same problem, no solution yet #617 There should be at least one working solution, however I couldn't find any so far

[gordinmitya]

overall onnx-tf doesn't seem working anymore I was able to convert mobilenet_v2 with https://github.com/nerox8664/pytorch2keras. Convert to saved_model then to tflite.