Deepstream RTSP Frames drop

AI & Data Science Computer Vision & Image Processing
,
1

Hi all ,
I am using a deepstream pipeline with multiple sources (RTSP*3) using “nvstreammux” with the following params
sync mode =1
max-latency = 1000000000
batched-push-timeout = 25000

But for some reason, sometimes i receive frames drop for one of the sources for approximately 30 seconds , may be it some buffer problem but i can not find proper property , I am familiar that “max-latency” property may effect that but this is not the case (at least in my opinion) , RTSP stream is working (can be seen in the player)

Is somebody familiar with other property that can be configured and help me with this scenario ?

Thanks in advance

2

Here are a few things you can check to resolve the issue:

  1. Network issues: Ensure that the network connection to the source is stable and not experiencing any latency or bandwidth issues. You can use network monitoring tools such as iperf or Wireshark to check the network conditions.
  2. CPU utilization: Make sure the CPU utilization is not reaching a high level, which could cause delays in the pipeline and result in dropped frames. You can use tools such as htop or top to monitor CPU utilization.
  3. GPU utilization: Similarly, check if the GPU utilization is high and causing the pipeline to slow down. You can use the nvidia-smi tool to monitor GPU utilization.
  4. Buffer size: Increasing the buffer size in the pipeline could help reduce the number of dropped frames. You can try adjusting the “nvbuf-memory-type” and “nvbuf-pool-size” properties in the pipeline to see if it helps.
  5. Gstreamer buffer properties: You can also try adjusting the Gstreamer buffer properties, such as “queue-size” and “leaky”, to see if it helps resolve the issue.
  6. Other sources: Ensure that the other sources in the pipeline are not causing any issues and contributing to the frame drop problem.