DeepStream SDK v1.5 in TX2 with Jetpack 3.3

AI & Data Science Deep Learning (Training & Inference)
1

Hi

I am having problem in running Deepstream sample code with Jetpack 3.3 release(Full update)

Executing below command gives segmentation fault and exits.

$ nvgstiva-app -c ${HOME}/configs/PGIE-FP16-CarType-CarMake-CarColor.txt

I am able to run the same with Jetpack 3.2(which is having TensorRT 3.0).

I want to verify the Yolo implementaiton which is not building with Jetpack 3.2. Throwing below error,

“IConstantLayer is not a member of nvinfer”

ALso w.r.t below link,

https://github.com/vat-nvidia/deepstream-plugins/tree/DS2

Yolo implementation requires CUDA 9.2 and TensorRT 4.x

Please help me in getting yolo implementation running with deepstream for TX2.

Thanks,
Partha

2

Please refer to the second point over here [url]https://github.com/vat-nvidia/deepstream-plugins/tree/7f27f22d674c64f4859dd1da896a43e35b0b9063#note[/url] to use the standard models with DS 1.5 and jetpack 3.3.

You cannot run the yolo app/plugin with Jetpack 3.2 since it requires TensorRT 4