The localization and tracking of users is an important part of sensing in smart environments. Superimposed intention recognition systems often rely on accurate user positions. In device-free localization systems (DFL) users do not need to wear any hardware in order to be located. In RF systems they can be localized using the changes in signal strength measured on static communication links. Recent work has shown that a combination of completely passive RFID hardware and a tomographic imaging approach can achieve a mean location error as low as 0.3 m in a post-processed validation. This work presents a system which is able to track the user in a live (online) mode with a constant frame rate while maintaining high accuracy. Conducting data from a demonstration setup the proposed system is able to locate users with a frame rate as low as 500 ms and a mean location error of 0.3 m.