Deploying tlt-1.0 models with deepstream 5.0 on jetpack4.4

Accelerated Computing Intelligent Video Analytics TAO Toolkit
1

I have upgraded my jetpack from 4.3 to 4.4 and Deepstream from 4.0 to 5.0. I am using my own custom trained models from TLT1.0. Can I directly deploy the model on deepstream 5.0 ? I don’t want to retrain my model using TLT 2.0. Also, are the models converted using tlt-converter for jetpack4.3 valid for jetpack4.4 ? or do I need to download the updated tlt-converter ?

3

Moving this topic from DS forum into TLT forum.

4
  1. You can try to deploy your old etlt model in DS.
  2. For trt engine you generated previously, it should not be working. If you run tlt-converter in Jetson platform, we support three kinds of tlt-converter.
    https://developer.nvidia.com/tlt-converter-trt51
    https://developer.nvidia.com/tlt-converter-trt60
    https://developer.nvidia.com/tlt-converter-trt71
5

Thanks.