Media-based modulation (MBM) is a novel modulation scheme, which embeds information in different channel fading realizations created by changing the status of radio frequency mirrors. In this letter, we investigate the problem of multi-user MBM signal detection in a large-scale multiple-input multiple-output system, where the base station is equipped with a large number of antennas and several users use MBM for uplink transmission. We propose an iterative interference cancellation-based algorithm. The basic idea is to treat the multi-user interference as noise and cancel it iteratively. Computer simulation results show that the proposed algorithm has good performance with low complexity.