The four video inputs become very delayed in deepstream-test3, why is that?

Accelerated Computing Intelligent Video Analytics DeepStream SDK
1
  • deepstream-app version 6.1.0
  • DeepStreamSDK 6.1.0
  • CUDA Driver Version: 11.4
  • CUDA Runtime Version: 11.0
  • TensorRT Version: 8.2
  • cuDNN Version: 8.4
  • libNVWarp360 Version: 2.0.1d3

Description：
When I use one video input fps can reach 60. The video in the detection process, it seems to be accelerated detection playback.
When I use four video inputs the fps is averaged to 15, the fps seems to be evenly split. At the same time, the video detection playback became delayed, as if it was slowed down.
I don’t think there should be this averaging, it should be a problem with the pipeline, right? I’m using deepstream-test3 and loading the engine model of yolov5.

3

did you do any modification? you can set batch-size to 4 if use four video inputs.

1 Like
4

I modified streammux’s batch-size=4 and the detected video became smooth. But the detection result is bouncing in different video windows

5

what does “the detection result is bouncing in different video windows” means? could you share a screenshot? what is your expecting?

6

Thanks for your reply, I have solved the problem by modifying the generated engine model.

1 Like
7

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