Tensorflow issue: no response after loading model

Autonomous Machines DRIVE AGX General
#1

Hi,

I’m trying to use Tensorflow on Xavier. All required libraries were installed and ran basic testing function is no problem.
But when we inferred the NN model, it stopped after loading model and no response, no error messages even if we wait 1 hours.
We ran the same code on X86 PC and it had no problem.

The model we used is VoxelNet and Deeplab v3 plus as below links.


Our software version:

  1. DriverWork 1.5
  2. tensorflow_gpu-1.13.0rc0+nv19.2-cp36-cp36m-linux_aarch64.whl

Does anyone can give me a hint to solve this problem?

Thanks.

tensorflow.png

#2

Dear wyang,
May I know why you want to install tensorflow on Drive AGX platform. Drive AGX platform is not intended for training and used for inference. If you are planning to do inference on Drive AGX platform, did you check our TensorRT APIs which can take tensorflow models and optimize the network furthur(https://docs.nvidia.com/deeplearning/sdk/tensorrt-developer-guide/index.html#import_tf_python).
For installing tesnorflow on Drive platform, could you please check https://devtalk.nvidia.com/default/topic/1044720/drive-agx/installing-tensorflow-from-source-on-drive-agx/post/5301621/#5301621