Python Samples of Deepstream 5.0 : Getting errors (Could not get EGL display connection) while running python test examples on Jetson Nano

Accelerated Computing Intelligent Video Analytics DeepStream SDK
22

Under /opt/nvidia/deepstream/deepstream-5.0, there uninstall.sh to uninstall the SDK.

23

As per FAQ mentioned here, uninstall.sh option is for dGPUs. For Nano or Xavier, do we need to re-flash the SD card with Jetpack4.4?

24

That suggested before, flash the device with deepstream option selected.

25

@Amycao, as you suggested, just to re-install DeepStream-5.0 we re-flashed the Jetson Nano SD Card and then installed DeepStream 5.0 with SDK manager.

The content of /usr/lib/aarch64-linux-gnu/gstreamer-1.0/deepstream is:

libgstnvvideoconvert.so    libnvdsgst_dsexample.so    libnvdsgst_msgbroker.so    libnvdsgst_multistreamtiler.so  libnvdsgst_osd.so
libnvdsgst_dewarper.so     libnvdsgst_inferserver.so  libnvdsgst_msgconv.so      libnvdsgst_of.so                libnvdsgst_segvisual.so
libnvdsgst_dsanalytics.so  libnvdsgst_infer.so        libnvdsgst_multistream.so  libnvdsgst_ofvisual.so          libnvdsgst_tracker.so

Now when I run C sample command -

./deepstream-app -c /opt/nvidia/deepstream/deepstream-5.0/samples/configs/deepstream-app/source8_1080p_dec_infer-resnet_tracker_tiled_display_fp16_nano.txt

I get same issue as earlier, when we tried DeepStream installation from command line using .deb file:

** ERROR: <create_multi_source_bin:1057>: Failed to create element 'src_bin_muxer'
** ERROR: <create_multi_source_bin:1132>: create_multi_source_bin failed
** ERROR: <create_pipeline:1296>: create_pipeline failed
** ERROR: <main:636>: Failed to create pipeline
Quitting
App run failed

This is strange just to re-install DeepStream, we have to re-flash the SD card with Jetpack 4.4.

Can you please help us to make DeepStream-5.0 on Nano and Xavier?

26

You installed Deepstream through Jetpack, right? see the picture,

1896041311
18960413111440×1080 126 KB

if you do like this, you should not meet the error.

27

@Amycao, yes we did same, selected DeepStream during installations.

28

You just reinstall deepstream, right? not a full fresh flash device and install software, right? if just reinstall deepstream, suggest to do a full fresh flash device and install software.

29

The steps I followed:

  1. I re-flashed the card from SD card image as mentioned here JetPack SDK | NVIDIA Developer
  2. Then on another Ubuntu machine, I installed SDK Manager
  3. Then we connected it to Nano and selected options like this:

SDK-First-screen
SDK-First-screen1400×900 240 KB

SDK-Screen-2
SDK-Screen-21400×900 229 KB

After this entered credentials and then it succeeded.

Let me know, if any other step/option is missing here.

30

Any help for this issue?

31

Please check this, make sure your image is same as Jetpack 4.4 equipped with.
nvidia@nvidia-desktop:/opt/nvidia/deepstream/deepstream-5.0$ cat /etc/nv_tegra_release |head
R32 (release), REVISION: 4.3, GCID: 21589087, BOARD: t186ref, EABI: aarch64, DATE: Fri Jun 26 04:34:27 UTC 2020

32

I got this output, which is same:

drishtic@drishtic-xavier:/opt/nvidia/deepstream/deepstream-5.0$ cat /etc/nv_tegra_release |head
R32 (release), REVISION: 4.3, GCID: 21589087, BOARD: t186ref, EABI: aarch64, DATE: Fri Jun 26 04:34:27 UTC 2020

33

Hi
Have you solved this issue?
After a fresh installation, you should not have the issue like mentioned.

34

@Amycao It was never succeeded from SDK manager GUI or command line. As mentioned here, it got resolved with oem-config only in one of our AGX Xavier from my host Desktop. On second one, Jetpack 4 succeeded, but DeepStream 5.0 still not working with SDK Manager installations with another host Desktop. Looks like I need to clean all existing SDK manager installations and downloaded files from host machine and try again. Something isn’t working with SDK Manager.