Computer Vision News - May 2024

25 Computer Vision News Zihan was a third-year bachelor’s student when NICE-SLAM was released, and he started working on NICER-SLAM while he was in his final year. In addition to making an impressive duo, Songyou and Zihan have worked with renowned researchers Andreas Geiger and Marc Pollefeys for several years. “They’re both very professional,” Zihan tells us. “They know a lot. They can think of papers from 20 years ago that directly relate to what they’re doing now. We presented something quite different from what they did during their PhD, but they understood it and could quickly point out their questions, thoughts, and possible improvements.” Songyou agrees: “They’re the smartest people I know. Andreas is super sharp and super organized. He gives good insights about your questions and how to solve something. He’s a very nice person as well and always encourages us. Marc is super smart and great at geometry or SLAM-related stuff, like structure from motion. He gives great feedback and provides a relaxed environment in which to explore different things under the big umbrella of 3D vision. He gives you the freedom to do whatever you want. That’s one of the reasons this work was possible.” NICER-SLAM arrives at a time when the field is rapidly advancing. Technologies such as Gaussian Splatting mark the next generation of NeRF, and many papers are trying to make it work in a similar SLAM setting. Yet, even in this fast-evolving landscape, NICER-SLAM stands out for its innovative approach to making SLAM more accessible and versatile, demonstrating robust and competitive performance against recent RGB-D systems. NICER-SLAM