Using Async code with DepthAI, is it possible?

lerp

Hey guys,

I've got OAK-1-Lite. Normally it can run "small" models like YOLO5n or YOLO8n on images that went through resizing to 416x416 or 640x640.

Question: Can I just make the code infer only every once in 60 seconds, so this way I could use larger models that take more inference time?

Moreoever, can I use Asynchoronus code? (*async)
*
Could I pass the inference to the VPU, using I/O logic of async, while the CPU continues to capture the frames and display them, until the coroutine of the inference function finishes?

thanks!

jakaskerl

lerp

lerp Question: Can I just make the code infer only every once in 60 seconds, so this way I could use larger models that take more inference time?

There is a limit to how large the model is. But it is possible to only run a single frame through the NN every 60 seconds. See https://docs.luxonis.com/software/depthai/examples/script_change_pipeline_flow/

lerp Moreoever, can I use Asynchoronus code? (*async)

RVC is asynchronous by nature, you don't need to do that.

Thanks,
Jaka