Deepstream sample application not working Jetpack4.4

.

**• Hardware Platform - Jetson Xavier NX
**• DeepStream Version - 5.0.1
**• JetPack Version - 4.4

I downloaded the Deepstream SDK from https://developer.nvidia.com/embedded/deepstream-on-jetson-downloads-archived
And I followed the methord2 and methord3 from Quickstart Guide — DeepStream 6.3 Release documentation to install deepstream sdk.
Deepstream is installed properly. On following https://github.com/NVIDIA-AI-IOT/deepstream_python_apps/blob/5cb4cb8be92e079acd07d911d265946580ea81cd/HOWTO.md , I cloned GitHub - NVIDIA-AI-IOT/deepstream_python_apps: DeepStream SDK Python bindings and sample applications
and ran

    $ cd /opt/nvidia/deepstream/deepstream-5.0/sources/deepstream_python_apps/apps/deepstream-test1 
    $ python3 deepstream_test_1.py /opt/nvidia/deepstream/deepstream-5.0/samples/streams/sample_720p.h264 

An error will then show “AttributeError: ‘NoneType’ object has no attribute ‘set_property’” because the “pgie” element (deepstream “nvinfer” element) is unable to be created properly. This is one of the blacklisted gstreamer plugins.

three plugins are shorlisted by gstreamer are:
libnvdsgst_infer.so
libnvdsgst_inferserver.so
libnvdsgst_osd.so

The reason for that being libnvrtc.so.10.2 cannot be opened.I checked the reason for that , cuda was not installed. For ubuntu 18.04(jetpack 4.4), it seems the cuda tool-kit version is CUDA 10.1 . I couldn’t find them on www.developer.nvidia.com/cuda-downloads/ . Where can I find the CUDA version 10.1 for deepstream 5.0 ?

Also , is there any specific installation steps of deepstream 5.0 ?

You need checkout on tag “v1.0” to support DeepStream 5.0: GitHub - NVIDIA-AI-IOT/deepstream_python_apps at v1.0

Please always upgrade to the latest DeepStream version if possible.

On checkout to v1.0 tag also , I see similar error . I am posting the error below

Playing file /opt/nvidia/deepstream/deepstream-5.0/samples/streams/sample_720p.h264
Traceback (most recent call last):
File “deepstream_test_1.py”, line 266, in
sys.exit(main(sys.argv))
File “deepstream_test_1.py”, line 203, in main
pgie.set_property(‘config-file-path’, “dstest1_pgie_config.txt”)
AttributeError: ‘NoneType’ object has no attribute ‘set_property’

And still the three files are blacklisted.
gst-inspect-1.0 -b
Blacklisted files:
libnvdsgst_inferserver.so
libnvdsgst_osd.so
libnvdsgst_infer.so

Total count: 3 blacklisted files

I couldn’t find an cuda installation candidate for ubuntu 18.0 (since jetpack 4.4 has ubuntu 18.0) on arm . Please help on that .

I’m afraid I need to upgrade jetpack first before installing DS-6.3.

Because DS-6.3 requires the latest version of CUDA and needs to update the kernel and driver.

You can use sdk manager

We are already working on jetpack 4.4 . But on SDK Manager , there is support to flash L4T staring from 4.6.2 Jetpack and above only. We use floyd board and update the package from jetson download package from jetson downloqad centre , archives section.

So , Kindly provide us a solution to run deepstream on Jetpack 4.4 itself

I think it is not a good idea to continue using jetpack 4.4, if you encounter some problems during development, it may not be supported on jetpack 4.4.

These problems are caused by Jetpack not being installed correctly. You can refer to the link below to re-burn Jetpack 4.4, and then refer to this link to install DeepStream-5.0. Finally, you can install deepstream python bindings.

https://docs.nvidia.com/metropolis/deepstream/5.0.1/dev-guide/text/DS_Quickstart.html

I face two issues,
#1 - In jetpack SDK 4.4 archive ( https://developer.nvidia.com/embedded/jetpack-sdk-44-archive ) , I find SDK Manager method . I downloaded the SDK Manager and tried , but it shows only from Jetpack 4.6.2 . Jetpack 4.4 is not available on SDK Manager.
#2 - I use diamond systems’s floyd developer kit , so I cannot directly flash the board with SDK manager . I have to download the packages from jetson download centre and addsome binaries on top of it and flash . So can you add steps on what I have to follow , to make deepstream working ,after I flash my floyd board with above mentioned method .

There is no update from you for a period, assuming this is not an issue anymore.
Hence we are closing this topic. If need further support, please open a new one.
Thanks

You can use SD Card Image Method and add some binaries after jetpack burning is completed.
The link I provided above has detailed steps.

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.