How do I insert a YOLOV5 inference module into deepstream in command-line

[hejun888]

hi,How do I insert a YOLOV5 inference engine into a deepstream as below

very thanks for this help

[marcoslucianops]

Replace path of primary inference to your config_infer_primary.txt path and execute your program with this command (with path to your libmyplugins.so file)

LD_PRELOAD=./libmyplugins.so ...

Where ... = your program command

[hejun888]

That doesn't seem to work,Is there any other way?Is it possible to place libmyplugins.so in config_infer_primary.txt like 'custom-lib-path=nvdsinfer_custom_impl_Yolo/libnvdsinfer_custom_impl_yolo.so' in config_infer_primary.txt

[marcoslucianops]

That doesn't seem to work,Is there any other way?

I know only this way for YOLOv5. Did you test like this?

LD_PRELOAD=./libmyplugins.so python3 deepstream.py

Is it possible to place libmyplugins.so in config_infer_primary.txt like 'custom-lib-path=nvdsinfer_custom_impl_Yolo/libnvdsinfer_custom_impl_yolo.so' in config_infer_primary.txt

Not for now, libnvdsinfer_custom_impl_yolo.so is a deepstream lib and libmyplugins.so is a tensorrtx lib. Need both libs to make YOLOv5 works.

[hejun888]

Can we compile two .so files(libmyplugins.so and libnvdsinfer_custom_impl_yolo.so ) into a single .so file in the following stages?

[marcoslucianops]

Can we compile two .so files(libmyplugins.so and libnvdsinfer_custom_impl_yolo.so ) into a single .so file in the following stages?

I will test

[hejun888]

hi,What's going on? Did the test work

[marcoslucianops]

hi,What's going on? Did the test work

Yes, I updated the repo, use my new files.

Now you can run with:

deepstream-app -c deepstream_app_config.txt

[hejun888]

thanks，it’s OK . Now it's easier than before.