Among other things, a person can ask the neural network what objects are captured in the frames by holding down the camera control button.