Old etlt models support in Deepstream 6+

Accelerated Computing Intelligent Video Analytics DeepStream SDK
1
  • Hardware Platform (Jetson/GPU): GPU
  • DeepStream version: 7.1

I have an ETLT model that was exported using an old TAO toolset. It runs on DeepStream 5.1. When I run the build engine from the ETLT model in DeepStream 7.1, the following error appears in the logs:
NvDsInferCudaEngineGetFromTltModel: UFF model support has been deprecated.

I have read that ETLT models exported with an older TAO toolkit can generate a UFF model, which is not supported in newer versions of DeepStream (6+).

Exporting the model using a new TAO toolkit compatible with DeepStream 7.1 would this new etlt be compatible with older versions of DeepStream?

Thanks for your help!

3

The etlt model is not supported any more with DeepStream 7.1. Can you tell us which model you want to deploy with DeepStream?

Please consult TAO toolkit forum for how to generate new ONNX model with corresponding TAO toolkit.

Latest Intelligent Video Analytics/TAO Toolkit topics - NVIDIA Developer Forums