Hi, I am using the ai_nvr workflow from Nvidia. I have ran ai_nvr peoplenet before and everything worked perfectly. I just did a fresh install of Jetson Linux on my Orin nano and now the deepstream is not doing inference on the camera source. The source does get detected and the stream shows perfectly on VST URL “http://192.168.101.110:30000/#/vst/live-streams“ and the source also appears on deepstream logs but still 0 active sources are shown. I have attached the logs here
*** Please note that this reply is generated by LLM automatically ***
Based on the provided context, it appears that you are experiencing issues with running the DeepStream application on your Jetson Orin Nano device after a fresh install of Jetson Linux. Specifically, you are not seeing any inference results on the camera source, despite the source being detected and the stream showing perfectly on the VST URL.
To troubleshoot this issue, let’s compare your setup with the recommended configuration for running DeepStream on Jetson Orin Nano.
JetPack version: Ensure that you are running the latest version of JetPack (6.1 GA or later) on your Jetson Orin Nano device. You can check the version by running the command dpkg -l | grep jetpack.
DeepStream version: Verify that you are using the latest version of DeepStream (6.1.1 or later) on your device. You can check the version by running the command dpkg -l | grep deepstream.
Camera configuration: Check that your camera is properly configured and recognized by the system. You can use the v4l2-ctl command to verify that the camera is detected and configured correctly.
DeepStream configuration: Review your DeepStream configuration file (e.g., deepstream_app_config.txt) to ensure that it is correctly set up for your camera source. Pay attention to the source and sink configurations, as well as the tracker and detector configurations.
Triton Inference Server: If you are using the Triton Inference Server, ensure that it is properly installed and configured on your device. You can check the status of the Triton server by running the command sudo systemctl status triton.
Comparing your setup with the recommended configuration, here are a few potential issues that might be causing the problem:
Incompatible JetPack version: If you are running an older version of JetPack, it might not be compatible with the latest version of DeepStream.
Incorrect camera configuration: If your camera is not properly configured or recognized by the system, DeepStream might not be able to detect and process the video feed.
Incorrect DeepStream configuration: If your DeepStream configuration file is not correctly set up for your camera source, it might not be able to process the video feed correctly.
To resolve the issue, you can try the following:
Update JetPack: Update your JetPack version to the latest available version (6.1 GA or later).
Verify camera configuration: Use the v4l2-ctl command to verify that your camera is detected and configured correctly.
Review DeepStream configuration: Review your DeepStream configuration file to ensure that it is correctly set up for your camera source.
Restart DeepStream: Restart the DeepStream application to ensure that it is using the correct configuration.
If none of these steps resolve the issue, please provide more details about your setup and configuration, including the exact error messages you are seeing in the DeepStream logs.
*** Please give the thumbs up if you get answers you like. Or provide feedback to help us improve the answer. ***
dpkg -l | grep jetpack
ii nvidia-jetpack 6.2.2+b24 arm64 NVIDIA Jetpack Meta Package
ii nvidia-jetpack-dev 6.2.2+b24 arm64 NVIDIA Jetpack dev Meta Package
ii nvidia-jetpack-runtime 6.2.2+b24 arm64 NVIDIA Jetpack runtime Meta Package
dpkg -l | grep deepstream
ii deepstream-7.1 7.1.0-1 arm64 Nvidia DeepStreamSDK runtime libraries, development files and samples
These are my jetpack and deepstream versions. Everything is installed properly I have verified it. ALso a new repo has been made by nvidia which is the l4t 36.5. Previously it was 36.4. Could that be causing an issue ?
I am using an RTSP camera which is perfectly working with the jetson. I have tested it using ffplay and also using VST. VST shows the stream of the camera. Even the deepstream container is detecting the proper name of the camera but its not doing inference on the camera because of a missing gstreamer plugin.
Opening in BLOCKING MODE
Error while setting IOCTL
Invalid control
S_EXT_CTRLS for CUDA_GPU_ID failed
nvmultiurisrcbin ERROR from element decodebin: Your GStreamer installation is missing a plug-in.
ERROR from decodebin: Your GStreamer installation is missing a plug-in.
Debug info: ../gst/playback/gstdecodebin2.c(4701): gst_decode_bin_expose (): /GstPipeline:deepstream-test5-0/GstDsNvMultiUriBin:multiurisrcbin/GstBin:multiurisrcbin_creator/GstDsNvUriSrcBin:dsnvurisrcbin0/GstDecodeBin:decodebin:
no suitable plugins found:
Couldn’t set nvv4l2decoder0 to PAUSED
nvmultiurisrcbin ERROR from element udpsrc1: Internal data stream error.
ERROR from udpsrc1: Internal data stream error.
Debug info: ../libs/gst/base/gstbasesrc.c(3127): gst_base_src_loop (): /GstPipeline:deepstream-test5-0/GstDsNvMultiUriBin:multiurisrcbin/GstBin:multiurisrcbin_creator/GstDsNvUriSrcBin:dsnvurisrcbin0/GstRTSPSrc:src/GstUDPSrc:udpsrc1:
streaming stopped, reason not-linked (-1)
I found above error in your log. Seems video decoder doesn’t work. The release note mentioned JPS tested on JP 6.1 GA (BSP R36.4.0) or JP 6.2 (BSP R36.4.3): Release Notes (version 2.0) — Jetson Platform Services documentation. Can you use it on JP 6.1 GA (BSP R36.4.0) or JP 6.2 (BSP R36.4.3)?