The Media Recognition and Understanding Research Group (MAGUS) focuses on the technological innovation and applications in computer vision and multimedia. We aim at discovering solutions to core problems in multimedia technology and improving machine perception for the benefit of users. In recent years, we have made some progress in the fields of visual media semantic understanding and environment perception for unmanned systems with papers published in high-ranked journals like MM, Tomm, and TMM, and have received multiple honors in the technical challenges hosted by influential conferences such as MM, ICME, and ECCV. Some of our research fundings have been transformed into industrial applications for leading companies including ATA, Samsung, and CSIC.