I am executing the RGB and tiny yolo example with my personal trained model, but the result gives me an extremelly low fps. There is any code line that allow me to raise the fps (like native opencv that i can use CUDA backend)?
Hi ThaidyDeguchi ,
AI inference is running on the device, which doesn't have opencv/cuda - it uses HW accelerated blocks specifically for ML models. Please check RVC2 NN performance to get a ballpark performance/FPS you can achieve. So for higher FPS I'd suggest lower resolution and/or smaller&lighter model.
I have an OAK-1 and with 1080p but I am achieving only 1 fps. There is a way to improve that based on this example: https://docs.luxonis.com/projects/api/en/latest/samples/Yolo/tiny_yolo/