Unable to build yolo plugin.

Accelerated Computing Intelligent Video Analytics DeepStream SDK
1

I am trying to get the Yolo Reference application to run on the deepstream docker container for Tesla (1080 ti)(https://github.com/NVIDIA-AI-IOT/deepstream_reference_apps/tree/master/yolo). While building the NvYolo plugin I get the following error:

Could NOT find CUDA (missing: CUDA_NVCC_EXECUTABLE CUDA_INCLUDE_DIRS
CUDA_CUDART_LIBRARY) (Required is exact version “10.0”)

I mentioned the path to cuda-10.0 as CUDA_TOOLKIT_ROOT_DIR.

This doesn’t turn up for deepstream-yolo-app. The cmake command executes successfully but make causes the error that
fatal error: gstnvdsmeta.h: No such file or Directory.

I had made the changes in Makefile.config by specifying Platform as tesla there. Am I doing something wrong? Or am I not doing something that’s required? Any help would be greatly appreciated.
Thanks for reading.

2

Hi,

Do you update the platform information in this configure file?
https://github.com/NVIDIA-AI-IOT/deepstream_reference_apps/blob/master/Makefile.config#L28

PLATFORM:=TESLA

Thanks.

3

Yes, i did that too. Didn’t seem to help.

4

The issue was resolved once I moved over to using a PC without docker. Apparently the CUDA-10.0 base on the docker did not have the required files. So, using a PC with cuda 10 and all required libraries worked fine for me.

5

Is there any solution to this using the docker image?

6

One thing you could try is using cuda-10 devel, if you’re facing the same issue. Might resolve this issue.