Ask Your Question

Tensorflow docker image not building PointPillars

asked 2019-11-04 04:27:36 -0600

Mackou gravatar image

Hello everyone,

  • Autoware version : 1.12.0
  • ROS Version Melodic
  • Autoware installed from Docker
  • Docker version 19.03.2, build 6a30dfc

I want to use PointPillars but it is not built when during the docker image build and I have this message in the log :

8942 Starting >>> lidar_shape_estimation

8943 ^[[91m--- stderr: lidar_point_pillars

8944 CUDA is available!

8945 CUDA Libs: /usr/local/cuda/lib64/libcudart_static.a;-lpthread;dl;/usr/lib/x86_64-linux-gnu/

8946 CUDA Headers: /usr/local/cuda/include

8947 TensorRT is NOT Available

8948 CUDNN is NOT Available

8949 PointPillars won't be built, CUDA and/or TensorRT were not found.

8950 ---

8951 ^[[0mFinished <<< lidar_point_pillars [5.09s]

So it seems that CUDNN and TensorRT are missing, Do you know if I missed something during the docker build or if I need to build autoware from source directly ?

Thanks a lot !

edit retag flag offensive close merge delete

1 Answer

Sort by ยป oldest newest most voted

answered 2019-11-11 17:13:14 -0600

Maximus5684 gravatar image

@Mackou The reason it is not building in our Docker image is because the CUDA image does not have CuDNN or TensorRT, only CUDA. You will have to create your own Docker image using ours as a base if you wish to add CuDNN and TensorRT. Building from source on your machine with these installed would be another alternative.

edit flag offensive delete link more

Your Answer

Please start posting anonymously - your entry will be published after you log in or create a new account.

Add Answer

Question Tools

1 follower


Asked: 2019-11-04 04:27:36 -0600

Seen: 29 times

Last updated: Nov 11