problem on making NV-YOLO plugin

Accelerated Computing Intelligent Video Analytics DeepStream SDK
1

I am trying to install the deepstream-yolo-app following the instructions in:
https://github.com/NVIDIA-AI-IOT/deepstream_reference_apps/blob/master/yolo/README.md

I receive the following error when I try to make the NV-YOLO plugin for Tesla

/usr/bin/ld: cannot find -lnvdsgst_helper
/usr/bin/ld: cannot find -lnvdsgst_meta
collect2: error: ld returned 1 exit status
CMakeFiles/gstnvyolo.dir/build.make:103: recipe for target 'libgstnvyolo.so' failed
make[2]: *** [libgstnvyolo.so] Error 1
CMakeFiles/Makefile2:67: recipe for target 'CMakeFiles/gstnvyolo.dir/all' failed
make[1]: *** [CMakeFiles/gstnvyolo.dir/all] Error 2
Makefile:129: recipe for target 'all' failed
make: *** [all] Error 2

I am using Deepstream SDK v3, TensorRT-5.0.2.6, Gstreamer v1.8.3

2

Hi,

After DeepStream installation, you should be able to find libnvdsgst_helper.so and libnvdsgst_meta.so in the /usr/local/deepstream/.
Could you help to check this first?

If you cannot find the library, please follow the README located at ${DeepStream_Release} to install SDK again.
Thanks.

3

I tried pulling the DeepStream docker image from NGC so the DS installation should be fine now. I am still experiencing problems since the docker’s CUDA version (10.1) does not match the expected CUDA version of the plugin (the make file depends exactly CUDA 10.0). Is there a way to use the nvyoloplugin in the DS docker image?
Moreover, Is there a docker image where both DS3 and TensorRT5 are built together?

Thanks

4

Hi,

Sorry that we don’t try this combination.
But there is no dependency between CUDA and yoloplugin.

You can update the Makefile based on your environment.
[url]https://github.com/NVIDIA-AI-IOT/deepstream_reference_apps/blob/master/Makefile.config#L26[/url]

Thanks.