A bird's-eye view of two papers released by Oh Hee-seok, professor in the AI Applications Department at Hansung University, and his research team. From left: conceptual diagrams of generative image editing and 3D point cloud quality assessment technologies. /Courtesy of Hansung University

Hanseong University said on the 26th that two papers released by Professor Oh Hee-seok's AI (artificial intelligence) applications lab were accepted to the international academic conference "CVPR 2026 (IEEE/CVF Conference on Computer Vision and Pattern Recognition)."

CVPR is a flagship academic stage where major universities and research institutes around the world and global corporations present their latest findings. It is considered the most prestigious conference in AI and computer vision.

The papers cover Generative AI-based natural image editing and 3D point cloud quality assessment. The university said, "It is meaningful in that achievements were made simultaneously in the core areas of 2D and 3D computer vision."

The first paper, "NEAF: Natural Image Editing with Attention Fusion for Generalizable Tuning-Free Text-Guided Image Editing," included Professor Oh and master's student Kim Ji-su. It is a zero-shot framework that performs image editing using text and image generation models without additional training. The university said it reflects user instructions while preserving the original image's structure and enables fast editing.

The second paper, , included Professor Oh; master's student Seo Sang-hyeok; and undergraduates Seo Jun-hyeok and Kim Da-un of the School of IT Convergence Engineering. This research is a no-reference framework that evaluates 3D point cloud quality without the original.

Professor Oh said, "This achievement demonstrates Hanseong University's research competitiveness in AI," adding, "We will continue to organically link education and research to produce sustained results."

※ This article has been translated by AI. Share your feedback here.