Question about YOLO's FPS and camera resolution

Hansi

Hi,
I'm a newbie to AI and I have a question about resolution and running speed.

When I ran YOLO on OAK-1, I found that YOLO's FPS dropped significantly when I increased the resolution of the camera. But as far as I know, the YOLO model will compress the size of the image frame to 416x416, so doesn't this mean that the size of the image frame of the running target detection should be the same regardless of the input image frame? But why does the FPS drop when I increase the resolution? Maybe OAK requires more operations to get higher resolution image frames, thus slowing down the overall speed?

Very much looking forward to your answer.

Best regards,
Hansi

erik

Hi Hansi ,
When you set model input size (eg. via tools.luxonis.com) this will set the while model size, so it won't do any resizing (eg. to 416x416). This will of course lead to lower speed (FPS). Thoughts?
Thanks, Erik