Hi, I am using DAK-1 for vehicle detection and tracking using YOLO. When I use the setup as 1080p, NN's fps can probably stabilize at 15fps, and that's good enough for my needs. But when I make more demands, for example, I want a higher resolution to 4k, this time the fps is only about 2 which is too low and I need at least 10. My purpose is to store pictures of each detected vehicle for later license plate detection and recognition, so 1080p resolution can't satisfy now, I need to increase the resolution to 4k. My question is can I set the resolution of the camera to 4k, then compress it to 1080p and input them to the neural network, then get the coordinate position of the vehicle, and later output the coordinate position to the original 4k frame and then crop it for storage but keep the fps still as 15 or at least better than 2? Thanks a lot for the answer.
