A Total of 23 Papers Accepted for Presentation at ‘ECCV 2026’
Yonhap News Agency
NAVER announced on the 28th that 23 of its research papers have been accepted for presentation at ‘ECCV (European Conference on Computer Vision) 2026,’ one of the world’s top conferences in the field of computer vision, thereby proving its competitiveness in spatial intelligence technology.
Celebrating its 19th edition this year, ECCV is considered one of the most prestigious AI conferences in video and image processing and is held biennially. The upcoming ECCV 2026 will be held in Sweden from September 8 to 12. Technical teams under Team NAVER, such as NAVER Labs, NAVER Labs Europe, NAVER AI Lab, and NAVER Cloud, will present their accepted papers at this event.
At this year’s conference, Team NAVER’s 23 papers were accepted as regular papers across various computer vision domains, including 3D spatial recognition and reconstruction, image and video learning models, and robotic driving simulation. In particular, the approval of a significant number of papers related to 3D spatial reconstruction has further validated NAVER’s technical prowess in the vision AI sector.
NAVER Labs Europe unveiled the follow-up study ‘BLASt3R’ to ‘DUSt3R,’ which previously garnered attention in the global 3D vision industry. BLASt3R integrates the ‘SLAM (Simultaneous Localization and Mapping)’ approach, enabling robots to generate maps of their surroundings in real time while on the move. This allows 3D reconstruction work to be carried out even in on-device environments.
NAVER Labs’ research on indoor 3D space creation drew academic attention by reconstructing room layouts and objects from a single 360-degree panoramic image. NAVER Labs also introduced technology that uses a VLM (Vision-Language Model) during autonomous driving to recognize previously unseen objects and accurately map them onto a real-time BEV (Bird’s-Eye View) map.
‘Seoul World Model,’ a research project conducted by NAVER AI Lab in partnership with KAIST, was accepted as a regular paper at the conference for its potential applications in map-based virtual exploration, digital twins, autonomous driving, and robotics simulation. Seoul World Model leverages NAVER’s vast spatial data, including panoramic images taken throughout the city, to generate long-distance videos that realistically depict Seoul.
NAVER Cloud revealed research achievements that further advanced AI’s ability to understand images and videos. The selected papers include research on accurately recognizing object movements within videos and studies that combine text-based inference with spatial images to better capture three-dimensional perceptions.
Hot Picks Today
“Excited for 400,000 Won” Turns to Anger: “It Feels Like an Insult” as Basic Pension Leaves Seniors with No Income Gain
Among these, a total of four papers—including research on 360-degree spatial reconstruction (InSpace) and Seoul World Model—were selected for the ‘ECCV Spotlight,’ which is awarded to about 1–2% of all submissions.
This content was produced with the assistance of AI translation services.
© The Asia Business Daily. All rights reserved. Unauthorized AI training and use prohibited.