Cannot convert ONNX model using tao_byom

[mayank.ukani]

Hello.

I have a yolo ONNX model which I have downloaded from here.

I have created Conda environment and everything as per steps mentioned in the documentation of tao_byom here. While trying to convert the model, it is throwing me error google.protobuf.message.DecodeError: Error parsing message with type 'onnx.ModelProto'

Please help understand/resolve this issue

• Hardware: X86 (RTX 3090 and AMD Threaripper)
• Network Type (Yolo_v3)
• How to reproduce the issue ? Just Download the any ONNX model from the link above and try to convert it using tao_byom. The issue will get reproduced

[Morganh]

Hi,
From BYOM Converter — TAO Toolkit 3.22.05 documentation
Currently, only the following models are supported by the TAO BYOM Converter:

• Classification
• UNet

[mayank.ukani]

So, if I understand correctly, we cannot directly convert any object detection model using tao byom and if I have any other format you guys are forcing the user to run the tao_converter on the same system where user want to deploy the model (source)

I have an ONNX model which I want to convert to an .plan file to deploy using ROS. If I use tao-converter it won’t let me convert as I have to provide an encoding key to convert. Which, I don’t have for this ONNX model.

How do you convert the model than ? I am confused.

Thanks for the quick reply
Mayank

[Morganh]

@mayank.ukani

we cannot directly convert any object detection model using tao byom

Yes, for BYOM, currently only the Unet and Classification models are supported by the TAO BYOM Converter.

The tool tao-converter is used to convert .tlt model into .etlt model.

If the .onnx model is not Unet or Classification, currently it is not workable to convert it to be a pretrained model to be running with TAO.

But you can build a TensorRT engine from ONNX using the trtexec tool . See Quick Start Guide :: NVIDIA Deep Learning TensorRT Documentation . But this is not a topic of TAO. It is a topic of TensorRT.