Conditionally change primary detection model - Deepstream 5.1

Hello everyone.

I am currently developing a processing pipeline for a project where I have different primary detectors depending on the conditions of the RTSP video input image. One model works when it receives visible RGB image and the other one when the video is IR imaging.

I have these two models and they both work correctly separately. However, I don’t know how to proceed to dynamically switch between one model and the other when the camera switches from visible to infrared and vice versa.

Is there any plugging or filter that can take care of choosing the appropriate model according to the conditions of the video input?

• Hardware Platform: Jetson & GPU (DOCKER)
• DeepStream Version: 5.1 (Python wrapper)

Thank you for your attention.
Juan Isern

Hi,
It looks similar to
Mux two pipelines after Demuxing - #7 by DaneLLL

For this it would require multiple nvstreammux plugins. We don’t support it on DeepStream SDK 5.1. We are evaluating to support it in future release.