Converting TF 2.0 saved model for TensorRT on Jetson Nano

Question

I am trying to convert a TF 2.0 saved_model to tensorRT on the Jetson Nano.

The model was saved in TF 2.0.0. The nano has Jetpack 4.2.2 w/ TensorRT __ and Tensorflow 1.14 (that is the latest Tensorflow release for Jetson).

I have been following the instuctions from here which describe how to convert a TF 2.0.0 saved_model into TensorRT.

Below is my code:

import tensorflow as tf
from tensorflow.python.compiler.tensorrt import trt_convert as trt
tf.enable_eager_execution()

converter = trt.TrtGraphConverterV2(input_saved_model_dir=input_saved_model_dir)
converter.convert()
converter.save(output_saved_model_dir)

saved_model_loaded = tf.saved_model.load(
    output_saved_model_dir, tags=[tag_constants.SERVING])
graph_func = saved_model_loaded.signatures[
    signature_constants.DEFAULT_SERVING_SIGNATURE_DEF_KEY]
frozen_func = convert_to_constants.convert_variables_to_constants_v2(
    graph_func)
def wrap_func(*args, **kwargs):
    # Assumes frozen_func has one output tensor
    return frozen_func(*args, **kwargs)[0]
output = wrap_func(input_data).numpy()

It seems to start converting successfully. However I get an KeyError: 'serving_default' error when it reaches the convert_to_tensor line. My complete printout is below found here (too long for SO), but the python traceback appears below. How can I fix this?

Thanks!

printout summary (complete printout here):

Traceback (most recent call last):
  File "tst.py", line 38, in <module>
    convert_savedmodel()
  File "tst.py", line 24, in convert_savedmodel
    converter.convert()
  File "/usr/local/lib/python3.6/dist-packages/tensorflow/python/compiler/tensorrt/trt_convert.py", line 956, in convert
    func = self._saved_model.signatures[self._input_saved_model_signature_key]
  File "/usr/local/lib/python3.6/dist-packages/tensorflow/python/saved_model/signature_serialization.py", line 196, in __getitem__
    return self._signatures[key]
KeyError: 'serving_default'

Answer 1

I can see two problems in your experiment:

• You are using TF-TRT 2.0 API while having TF 1.14 installed. That is not supported. If you have TF 1.14 installed on your system, then you would need to use TF-TRT 1.x API.

• TF Models saved in TF2.0 are not compatible with TF1.14 according to https://www.tensorflow.org/guide/versions

If you only have access to TF1.14, I suggest to re-generate the graph in TF1.14 and save the model there before applying TF-TRT, and then use TF-TRT 1.x API.