Use deep stream for custom applications

Please provide complete information as applicable to your setup.

• Hardware Platform (Jetson / GPU) : Jetson
• DeepStream Version 5.0 DP
• JetPack Version (valid for Jetson only) 4.4 dev kit
• TensorRT Version :

Hi guys,
In your opinion, Is it possible to use deep stream for custom app? for example : face recognition
That’s mean I want to use decode multi-stream and one detector of deep stream for face detection, but for face recognition, the deep stream doesn’t support any model for this task, I want to know How I can integrated the face rec system to deep steam? I want to get outputs like counting and coordinates of deep stream, Is it possible?


Face recognition can be understood as a classification task.
DeepStream pipeline supports detection + classification scenarios.

You can find a sample of such pipeline in <where_ds_is_installed>/deepstream-5.0/sources/apps/sample_apps/deepstream-test2

You right, But for face recognition I need also search in database for persons, I need to get similarity for feeded images in database, for this problem How do I do? How I can to add the search of persons codes into deepstream system?


You can do that. But DS pipeline is data buffer driven so it is susceptible to any disturbance of other jobs & executions if they are synchronized. You should try to make these extra operations asynchronized to minimize negative effects to performances.

You mean is that I need to write custom gstreamer plugin for extra operation and then I need to add into pipeline? after face rec(classification) plugin? If so, How I can to see what’s the classification outputs in pipeline?

and in the gstreamer concept every plugin has srouce and sink pads, right? sink pad is input or output of plugin?


I did not mean you should add a new gsteamer plugin to do this. I mean any extra operation like file & network reading/writing or database update/query may affect pipeline performance if these operations are synchronized. Any async implementation is okay.