Error when run example NVDECINFER_DETECTION

I have downloaded DeepStrean 1.5 and compile all given examples. All examples are compiled fine but when I try to run the NVDECINFER_DETECTION example, it fails after reading the model’s description:

[DEBUG][17:13:28] Device ID for display [0]: Tesla P4
[DEBUG][17:13:28] Device ID for inference [0]: Tesla P4
[DEBUG][17:13:28] Video channels: 4
[ERROR][17:13:28] Warning: No mean files.
[DEBUG][17:13:28] GUI enabled.
[DEBUG][17:13:28] Endless Loop: 0
[DEBUG][17:13:28] Device name: Tesla P4
[DEBUG][17:13:29] Use INT8 data type.
[DEBUG][17:13:31] =========== Network Parameters Begin ===========
[DEBUG][17:13:31] Network Input:
[DEBUG][17:13:31]       >Batch   :4
[DEBUG][17:13:31]       >Channel :3
[DEBUG][17:13:31]       >Height  :368
[DEBUG][17:13:31]       >Width   :640
[DEBUG][17:13:31] Network Output [0]
[DEBUG][17:13:31]       >Channel :4
[DEBUG][17:13:31]       >Height  :23
[DEBUG][17:13:31]       >Width   :40
[DEBUG][17:13:31] Network Output [1]
[DEBUG][17:13:31]       >Channel :16
[DEBUG][17:13:31]       >Height  :23
[DEBUG][17:13:31]       >Width   :40
[DEBUG][17:13:31] =========== Network Parameters End   ===========
freeglut (dummy): failed to open display ''

I am using ssh to access Ubuntu 16.04 LTS on a P4 server

Hi,

Could you try this command and test it again?

export DISPLAY=:0

Thanks.

I tried this command but it still failed.

freeglut (dummy): failed to open display ':0'

My server is using two Xeon processors with only one P4

Hi,

Do you use display driver 384?
Could you share your environment information with us? Including OS version, CUDA version and display driver version.

Thanks.

I am using driver 390.30 on Ubuntu 16.04.4 LTS with CUDA 9.0.176

Thank you for your help. I have edited the run.sh file and found that the demo was made to be run on a Titan X GPU, not a P4 as mentioned in the document as well as the gui option was set to 1 as default while in the document, NVIDIA said that it was set to 0 as default. I hope in the next release these conflicts should be resolved.

Hi,

Thanks for your feedback.
We will update your comment to our internal team.

Thanks.