No importer registered for op: Gather ERROR: failed to parse onnx file

Autonomous Machines Jetson & Embedded Systems Jetson TX2
1

When i run the tensorrt/samples/samplesONNXMNIST/samplesONNXMNIST.cpp. I meet this errors:

Input filename: …/data/mnist/resnet18.onnx
ONNX IR version: 0.0.4
Opset version: 9
Producer name: pytorch
Producer version: 1.1
Domain:
Model version: 0
Doc string:

WARNING: ONNX model has a newer ir_version (0.0.4) than this parser was built against (0.0.3).
While parsing node number 69 [Gather → “192”]:
ERROR: /home/erisuser/p4sw/sw/gpgpu/MachineLearning/DIT/release/5.0/parsers/onnxOpenSource/ModelImporter.cpp:142 In function importNode:
[8] No importer registered for op: Gather
ERROR: failed to parse onnx file

I was transferring an onnx model to tensorrt. The onnx model comes from pytorch pretrained model. I tried standard resnet50, and use The torch.onnx.export() function. When parsing this generated onnx model, it always appears the above error.but i use the mnist.onnx which is the offical model.it is ok.
Besides,I run it on tx2. but tx2 the lastest version just support tensorRT5.0. I can not install tensorRT5.1 on tx2.Because these files are compiled for x86-64 architecture not arm64 which TX2 used. Please help me. Thanks first!

2

Hi,

ONNX parser support Gather until TensorRT5.1.
Please wait for our next JetPack release to get the layer support.
You can find our future schedule on this topic:
https://devtalk.nvidia.com/default/topic/1055628/jetson-nano/new-jetson-software-modules-and-pricing/

Alternatively, maybe you can try to compile a v5.1 ONNX parser for Jetson.
But it may have some dependency issue between parser(v5.1) and inference(v5.0) library.
https://github.com/onnx/onnx-tensorrt/tree/8b52755beb781366f3bd2a34b3a0be0fcd4c0c33

Thanks.

3

Hi,

JetPack4.2.1 just release now.
There is a TensorRT5.1 for Jetson platform now.

Thanks.