As it turns out using services for the torch models was a bad idea, previously it was taking 0.2-10 seconds to get my torch info back. But now its like 0.05-0.2 seconds which is reasonable for what I am doing (also this is over the ekf filtering which is added in this pr).
As it turns out using services for the torch models was a bad idea, previously it was taking 0.2-10 seconds to get my torch info back. But now its like 0.05-0.2 seconds which is reasonable for what I am doing (also this is over the ekf filtering which is added in this pr).