Introduction to the Scalable Graphics, Vision & Robotics Lab
SGVR Lab covers research areas of computer graphics, motion planning, and image retrieval. The objective of our research is to design and process efficient and scalable visual data. The more complex data with high quality is generated thanks to prevalence of smart devices, the more important techniques to handle them rapidly become. To address those issues, SGVR Lab is mainly interested in the following three areas: （1）scalable rendering, （2）motion planning, and （3）image retrieval.
We study frameworks using CPU/GPU for massive model rendering in real time, and filtering methods for photo-realistic rendering.
Figure 4 We have developed high-quality rendering systems that can render the Boing 777 model interactively.
We study path generation of robotics to sense surrounding environments, avoid obstacles, control without human-aided. We also focus on optimization, convergence, and uncertainty of path generation.
Figure 5 We have participated in the KAIST autonomous vehicle team, where we have developed motion planning techniques.
We study how to search similar images in large-scale database, including millions or billions of images. The main topics are compression of visual data and similarity computation for retrieval.
Figure 6 An image search system that we have collaborated with Adobe, USA.