I am working on multistream detection using DeepStream. I trained a YOLOv8 model and successfully ran it with 8 streams. However, I am uncertain about the optimal configuration for the streammux parameters.
The streams have a resolution of 1920x1080, while the YOLOv8 model was trained with an input resolution of 640x640.
To achieve the best performance, what should the width and height parameters in the streammux section be set to? Should they match the stream resolution (1920x1080) or the model’s input resolution (640x640)? Additionally, should I enable padding by setting enable-padding=1?
Here is my current configuration for streammux:
[streammux]
gpu-id=0
live-source=0
batch-size=8
batched-push-timeout=40000
width=1920
height=1080
enable-padding=0
nvbuf-memory-type=0