Unable to compile deepstream LPR application

Accelerated Computing Intelligent Video Analytics DeepStream SDK
, ,
1

• Hardware Platform: HPE Proliant-DL-385-GEN-10-Plus
• Graphics Card : Tesla T4
• OS : Ubuntu 18.04 (64-bit)
• NVIDIA GPU Driver Version : 450.102.04
• Cuda Version : 10.2
• cuDNN Version: 8.0.2.39
• TensorRT Version : 7.1.3.4
• DeepStream Version : 5.0

• Issue Type: Unable to do generate libnvdsinfer_custom_impl_lpr.so file
• How to reproduce the issue ?
Reproduce:
Step-1: Clone the deepstream_lpr_app from git repository inside /opt/nvidia/deepstream/deepstream-5.0/sources
Step-2: cd deepstream_lpr_app/nvinfer_custom_lpr_parser → make

After giving make command the following error occurs:
g++ -o libnvdsinfer_custom_impl_lpr.so nvinfer_custom_lpr_parser.cpp -Wall -Werror -std=c++11 -shared -fPIC -Wno-error=deprecated-declarations -I/opt/nvidia/deepstream/deepstream-5.0/sources/includes -Wl,–start-group -lnvinfer -lnvparsers -Wl,–end-group
/usr/bin/ld: cannot find -lnvinfer
/usr/bin/ld: cannot find -lnvparsers
collect2: error: ld returned 1 exit status
Makefile:37: recipe for target ‘libnvdsinfer_custom_impl_lpr.so’ failed
make: *** [libnvdsinfer_custom_impl_lpr.so] Error 1

• Requirement details I am working on this machine for doing the LPR model training…

Can someone help to solve this issue

3

Seems you didn’t install the TensorRT correctly, you can check if libnvinfer.so inside your systerm lib path.

4

:~/Downloads/TensorRT-7.1.3.4/lib$ ls
libmyelin_compiler_static.a libnvinfer_plugin.so libnvonnxparser_static.a
libmyelin_executor_static.a libnvinfer_plugin.so.7 libnvparsers.so
libmyelin_pattern_library_static.a libnvinfer_plugin.so.7.1.3 libnvparsers.so.7
libmyelin_pattern_runtime_static.a libnvinfer_plugin_static.a libnvparsers.so.7.1.3
libmyelin.so libnvinfer.so libnvparsers_static.a
libmyelin.so.1 libnvinfer.so.7 libonnx_proto.a
libmyelin.so.1.1.0 libnvinfer.so.7.1.3 libprotobuf.a
libnvcaffe_parser.a libnvinfer_static.a libprotobuf-lite.a
libnvcaffe_parser.so libnvonnxparser.so stubs
libnvcaffe_parser.so.7 libnvonnxparser.so.7
libnvcaffe_parser.so.7.1.3 libnvonnxparser.so.7.1.3

Im having

5

That’s the problem, you didn’t install these libs in your system lib path, so compiler cannot find it.
It’s just a basic compiler problem.
2 options to fix it:
Add -L to specify the search path for compiler
install these lib in your system lib path

6

Attach Using and Porting the GNU Compiler Collection (GCC): Invoking GCC for your reference

7

I have installed TensorRT by following this link only

8

I got solution for this issue.

  1. I have installed cuda-10.2 as run file method-> so I uninstalled cuda and re-installed cuda-10.2 as deb method
  2. Removed the packages of existing cudnn libraries which was not removed properly
  3. Uninstalled TensorRT because I have installed it using tar file method → then I re-installed it using deb method

After I did the above changes I’m able to compile the application

Thanks for the assistance