Retrained etlt to DeepStream Example

Please provide complete information as applicable to your setup.

• Hardware Platform (Jetson / GPU) Jetson Xavier AGX
• DeepStream Version 4.0
• JetPack Version (valid for Jetson only) 4.3
• TensorRT Version 6.0.1
• NVIDIA GPU Driver Version (valid for GPU only)

I have trained a sample Faster-RCNN in TLT and exported it as .etlt. Then, I was trying to use in the DeepStream. However, the custom plugin required for the F-RCNN in TLT and the samples (which are uff) in DeepStream are different.

I have also tried the etlt version from anothe sample but it seems that it is not the same plugin required as the sample from TLT as seens in this link https://github.com/NVIDIA-AI-IOT/deepstream_4.x_apps.

The TLT mentioned that we need cropAndResize and ProposalPlugin. I have compiled the OSS for TensorRT 6.0. I changed the custom plugin to libinvinfer_plugin.so but it failed. If there is any example for such scenario ? Thanks.

Hi cpchiu,
What’s the error when you mention “I changed the custom plugin to libinvinfer_plugin.so but it failed.”?
For you case, please refer to “Option 2” of https://github.com/NVIDIA-AI-IOT/deepstream_4.x_apps/tree/release/1.0.1