Tao-toolkit-triton-apps using peoplesemsegnet model

[clairewang0409]

Hi

I have a problem that the inference results from tao unet jupyter notebook and from
tao-toolkit-triton-apps are different but using the same model.

Therefore, I tried to use the peoplesemsegnet model form your website to check the problem is related to inference platform or the model.

However, I got an error when I tried to using the peoplesemsegnet model on tao-toolkit-triton-apps.

Below are my command & error log:

python tao_client.py ./test3 -m test3 --mode Test -i https -u localhost:8000 --async --output_path ~/tao-toolkit-triton-apps/tao_triton/python/entrypoints/results3

Traceback (most recent call last):
File “tao_client.py”, line 447, in
main()
File “tao_client.py”, line 428, in main
responses.append(async_request.get_result())
File “/home/justin927/anaconda3/envs/deepstream/lib/python3.6/site-packages/tritonclient/http/init.py”, line 1590, in get_result
_raise_if_error(response)
File “/home/justin927/anaconda3/envs/deepstream/lib/python3.6/site-packages/tritonclient/http/init.py”, line 64, in _raise_if_error
raise error

The attachment is my config file. Because I didn’t know how to set the config, so I just used my unet model setting. If there are any config setting reference?

config.pbtxt (246 Bytes)

[Morganh]

Could you double check if follow the steps in GitHub - NVIDIA-AI-IOT/tao-toolkit-triton-apps: Sample app code for deploying TAO Toolkit trained models to Triton?

[clairewang0409]

The sample models and my own unet model can run successfully.
I think this problem is because I don’t know how to set the config file of your peoplesemsegnet model.
So, I want to know if there are any config setting reference for triton of peoplesemsegnet?
Thanks!

[Morganh]

There is no update from you for a period, assuming this is not an issue anymore. Hence we are closing this topic. If need further support, please open a new one. Thanks

Unfortunately, currently unet(or peoplesemsegnet) inference is not supported in tao-toolkit-triton-apps.
Please change to
https://docs.nvidia.com/tao/tao-toolkit/text/ds_tao/deepstream_tao_integration.html#peoplesemsegnet
or actually GitHub - NVIDIA-AI-IOT/deepstream_tao_apps: Sample apps to demonstrate how to deploy models trained with TAO on DeepStream to use peoplesemsegnet model.