On screen display frame drop

Accelerated Computing Intelligent Video Analytics DeepStream SDK
1

Hello,
I am currently working with DeepStream on a Jetson Nano with DeepStream 5.0 and jetpack 4.4.
I used the deepstream-test1 application and created a new application taking as an input the CSI camera.
After this modification, I noticed that the on screen display is dropping frames and only displays the camera image with the detected object at about 1 FPS.

On another try, I took deepstream-test1 and modified the neural network to PeopleNet that runs at 10 FPS on the Nano Jetson. However, the on screen display is also very very slow.

Do you know what could cause this?
Thanks for your help,
Clément

3

Hi,
Please apply the change

and try again.

1 Like
4

It works great! Only minor detail is that I can’t close the video after starting the program. Do you know the key combination to close it?

Thanks again,

5

Hi,

Please send EoS:

    GstElement* src = gst_bin_get_by_name(GST_BIN(gst_pipeline), "mysource");
    gst_element_send_event (src, gst_event_new_eos ());
    // Wait for EOS message
    GstBus *bus = gst_pipeline_get_bus(GST_PIPELINE(gst_pipeline));
    gst_bus_poll(bus, GST_MESSAGE_EOS, GST_CLOCK_TIME_NONE);

A reference sample:

6

That works well. Would it be possible to have it not in full screen but in windowed mode?

Thanks again,

7

Hi,
The nveglglessink plugin is window-based sink. Please execute sudo jetson_clocks and try nveglglessink sync=0.