3DiffGS accepted to BMVC 2026
Our paper, ‘3DiffGS: 3D Gaussian Splats from Unposed 2D Images using Diffusion Models,’ has been accepted to BMVC 2026. Congratulations to Abhigyan and the entire NAAMII team behind this project!
Updates / SIL
News
Publications, recognition, people, and other notes from the Spatial Intelligence Lab.
ECCV 2026 paper selected for a Long Oral
Our paper, ‘Seeing Through the Weights: Privacy Leakage in Scene Coordinate Regression,’ has been selected for a Long Oral presentation at ECCV 2026, placing it among approximately the top 2% of submissions. Congratulations to Oleksii Nasypanyi, Jaemin Cho, Utku Ozbulak, Bohyung Kang, and François Rameau!
Read more ↗
Privacy-preserving visual localization accepted to ECCV 2026
Our work examining privacy leakage in scene-coordinate regression has been accepted to ECCV 2026. Congratulations to Oleksii Nasypanyi and Jaemin Cho.
Read more ↗Outstanding Reviewer at CVPR 2026
Prof. François Rameau was recognized among the Outstanding Reviewers for CVPR 2026 for rigorous and constructive service to the computer-vision community.
Read more ↗
Invited talk at the SUNY AI Symposium
Prof. Rameau presented an overview of the AI research landscape in Korea at the SUNY AI Symposium in Stony Brook and joined discussions with researchers and participants.
Read more ↗
Domain adaptation work accepted to ICPR 2026
Our work on dual-foundation models for unsupervised domain adaptation has been accepted to ICPR 2026.
Read more ↗
Spurious-correlation work accepted to TMLR
Our token-based approach for detecting spurious correlations in vision transformers has been accepted to Transactions on Machine Learning Research, led in collaboration with the Ghent University team.
Read more ↗
Principal Chair at ANAIS 2025
The Annual Nepal AI School welcomed 225 participants from 19 countries for an intensive 11-day program with 23 speakers. Prof. Rameau served as Principal Chair with Ashutosh Modi and Vijay Prakash Dwivedi.
Read more ↗
Panel discussion on AI risk in Kathmandu
Prof. Rameau joined banking leaders, academic experts, and public officials to discuss AI risks, institutional readiness, and leadership challenges in the financial sector.
Read more ↗
Best Paper Award at ICONIP 2025
Our work on practical 16-bit neural-network training received the Best Paper Award at ICONIP 2025.
Read more ↗
SUNY Korea Presidential Award
Jaemin Cho and Oleksii Nasypanyi received the SUNY Korea Presidential Award for their research on privacy attacks in scene-coordinate regression.
Read more ↗
Epipolar-guided matching accepted to 3DV 2026
Our work on pixel-accurate epipolar-guided matching has been accepted to the International Conference on 3D Vision.
Read more ↗