Use tlt-converter to get DLA .engine from .etlt

Accelerated Computing Intelligent Video Analytics TAO Toolkit
1

I have download tlt-converter on Jetson Xavier NX. I have the PeopleNet pruned quantized .etlt (https://ngc.nvidia.com/catalog/models/nvidia:tlt_peoplenet/files?version=pruned_quantized_v2.1) and int8 calib file. I want to make an engine for DLA. How to?

Update: Solved, I had to use the DLA specific version of tlt-converter from here https://developer.nvidia.com/assets/TLT/Secure/tlt-converter-7.1-dla.zip

3

Please refer to How to run tlt-converter

Note, please download tlt-converter with dla enabled. wget https://developer.nvidia.com/assets/TLT/Secure/tlt-converter-7.1-dla.zip

Then set “-u 0” or " -u 1".
Reference: DetectNet_v2 — Transfer Learning Toolkit 3.0 documentation

1 Like