Deepstream python sample apps can just run at 30fps when showing display on orin

Accelerated Computing Intelligent Video Analytics DeepStream SDK
1

Please provide complete information as applicable to your setup.

• Hardware Platform (Jetson / GPU)
Jetson Orin
• DeepStream Version
6.2
• JetPack Version (valid for Jetson only)
5.0.2
• TensorRT Version
8.4.1-1+cuda11.4

Hello, I have been working with the deepstream python sample apps with my custom yolov4-tiny models.
I used deepstream-test3 and deepstream-imagedata-multistream-redaction that outputs a Rtsp stream.

Both projetcs do 30 FPS, when I hide the display in deepstream-test3 using ‘–no-display’ I can get up to 170 fps.
Using deepstream-imagedata-multistream-redaction I get 30 FPS in the Rtsp stream.
I have also tested both projects using the sample models from nvidia and I get the same results.

Is this correct?
Is 30 FPS the maximun that deepstream can handle when showing the display?

3

Have you measured your custom yolov4-tiny model with “trtexec”? Have you followed the instruction of performance testing here Performance — DeepStream documentation

We don’t know. It depends on your settings and configurations.

Depends on your settings, configurations or even the network performance if you are using RTSP. The FPS is for the whole application but not only for GPU.

5

trtexec gives me

image
image963×444 97.4 KB

6

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

Seems the model is not a bottleneck. You can test the performance according to Performance — DeepStream 6.1.1 Release documentation (nvidia.com)