SEOUL, February 27 (AJP) - A research team led by Professor Lee Seong-won of the Department of Electronics Engineering at Kookmin University has had two papers accepted for the IEEE/CVF Conference on Computer Vision and Pattern Recognition 2026.
The IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) is regarded as a premier international academic event in the fields of artificial intelligence and computer vision. Known for its rigorous review process and low acceptance rates, the conference serves as a global stage where leading universities and research institutions share core technological advancements.
The research was conducted through a joint collaboration between the team of Professor Lee Seong-won at Kookmin University and the team of Professor Kim Eun-tae at Yonsei University. The two institutions worked together to develop methods that combine theoretical precision with practical scalability.
The first paper, titled Rethinking Pose Refinement in 3D Gaussian Splatting under Pose Prior and Geometric Uncertainty, proposes a new optimization framework to address camera pose errors and geometric uncertainties. These issues are central challenges in 3D Gaussian Splatting, a prominent technique for 3D scene reconstruction. By integrating pose priors and geometric uncertainty analysis, the framework enables more stable and precise three-dimensional reconstruction, with potential applications in digital twins, robotics, and immersive content creation.
The second paper, titled HypeVPR: Exploring Hyperbolic Space for Perspective to Equirectangular Visual Place Recognition, introduces a method for visual place recognition using hyperbolic space. This approach is designed to overcome representation differences between standard perspective images and equirectangular (omnidirectional) images. The framework maintains robust recognition performance across different camera formats, providing a foundational technology for autonomous driving, drones, and extended reality (XR).
CVPR is a conference where computer vision researchers worldwide aspire to present, as it strictly evaluates the originality and completeness of research, Professor Lee Seong-won said. We will continue to strengthen our global competitiveness through research that possesses both theoretical sophistication and practical application.
Professor Lee Seong-won added that Kookmin University (KU) intends to focus on creating leading research results and fostering talent in the fields of artificial intelligence and computer vision.
Copyright ⓒ Aju Press All rights reserved.