Failed to convert onnx to trt build_engine.py

AI & Data Science Deep Learning (Training & Inference) TensorRT
#1

Description

Failed to build engine.py on the Jetson Nano jetpack 4.6
I build the onnx on my laptop using command python -m tf2onnx.convert --saved-model H:\Jupyter_Notebooks\ObjectDetection\workspace\training_demo\exported-models\my_model\saved_model --output H:\Jupyter_Notebooks\ObjectDetection\workspace\training_demo\onnx\model.onnx --opset 11
When trying to build engine at the nano this error happens. I can’t use opset 9 for my desktop and i can’t convert the model to onnx at the nano …

python build_engine.py --onnx onnx/model.onnx --engine engine/engine.trt --precision fp16
[TensorRT] INFO: [MemUsageChange] Init CUDA: CPU +198, GPU +0, now: CPU 228, GPU 3457 (MiB)
[TensorRT] WARNING: onnx2trt_utils.cpp:364: Your ONNX model has been generated with INT64 weights, while TensorRT does not natively support INT64. Attempting to cast down to INT32.
[TensorRT] INFO: No importer registered for op: TensorListStack. Attempting to import as plugin.
[TensorRT] INFO: Searching for plugin: TensorListStack, plugin_version: 1, plugin_namespace:
[TensorRT] ERROR: 3: getPluginCreator could not find plugin: TensorListStack version: 1
ERROR:EngineBuilder:Failed to load ONNX file: /home/rabah/TensorRT/samples/python/tensorflow_object_detection_api/onnx/model.onnx
ERROR:EngineBuilder:In node 9 (importFallbackPluginImporter): UNSUPPORTED_NODE: Assertion failed: creator && "Plugin not found, are the plugin name, version, and namespace c

Environment

TensorRT Version:
GPU Type:
Nvidia Driver Version:
CUDA Version:
CUDNN Version:
Operating System + Version:
Python Version (if applicable):
TensorFlow Version (if applicable):
PyTorch Version (if applicable):
Baremetal or Container (if container which image + tag):

Relevant Files

Please attach or include links to any models, data, files, or scripts necessary to reproduce your issue. (Github repo, Google Drive, Dropbox, etc.)

Steps To Reproduce

Please include:

  • Exact steps/commands to build your repro
  • Exact steps/commands to run your repro
  • Full traceback of errors encountered
#2

Hi,
Request you to share the ONNX model and the script if not shared already so that we can assist you better.
Alongside you can try few things:

  1. validating your model with the below snippet

check_model.py

import sys
import onnx
filename = yourONNXmodel
model = onnx.load(filename)
onnx.checker.check_model(model).
2) Try running your model with trtexec command.

In case you are still facing issue, request you to share the trtexec “”–verbose"" log for further debugging
Thanks!

#4

model.onnx (10.4 MB)

When I used create_onnx.py TensorRT Object Detection it always gave me AttributeError: 'Graph' object has no attribute 'op_with_const'
when I used tf2onnx.convert w/ default opset “9” it gave me this error raise ValueError("StridedSlice: attribute " + attr_name + " not supported") ValueError: StridedSlice: attribute new_axis_mask not supported
But when using opset 11 it succeeded
However, while validiting the model using the check_model.py it returned this error: `ValidationError: No Op registered for TensorListStack with domain_version of 11

==> Context: Bad node spec for node. Name: StatefulPartitionedCall/map/TensorArrayV2Stack_1/TensorListStack OpType: TensorListStack`

And as far as I know, Opset 9 is needed to work on the jetson …

#5

Hi,

Sorry, could you please give more details.
Were you able to successfully convert TensorFlow → ONNX-> TensorRT on opset 11?
Are you facing the above issue using opset 9 ?
Which version of the TensorRT are you using ?

#6

Sorry for the late reply but you took a while to answer so I forgot about that post. Actually yeah I did successfully convert tensorflow to onnx using opset 11 only. When using opset 9 ("StridedSlice: attribute " + attr_name + " not supported") ValueError: StridedSlice: attribute new_axis_mask not supported.
On my pc it’s TensorRT-8.2.4.2
My nano is JP43