RobustGenBench: A Benchmark for Robust Generalization to Adversarial and Common Perturbations, with Applications to Vision and Vision-Enabled Large Language Models

Maxime Heuillet, JONAS NGNAWE, Yann Pequignot, Rishika Bhagwatkar and 5 more

Sydney Poster Session 5, Thu, Dec 10, 10:00 AM–1:00 PM, Hall 1-4 · Published 2026

SyncLight: Single-Edit Multi-View Relighting

SyncLight propagates single-view lighting edits across arbitrary uncalibrated multi-view captures via a latent bridge-matched diffusion transformer, enabling consistent high-fidelity relighting without camera poses.

David Serrano-Lozano, Anand Bhattad, Luis Herranz, Jean-Francois Lalonde and 1 more

Paris Poster Session 5, Fri, Dec 11, 11:30 AM–1:30 PM, Paris Poster Hall · Published 2026

CanViT: Toward Active-Vision Foundation Models

CanViT introduces the first active-vision foundation model with a retinotopic backbone and scene-wide canvas, achieving 38.5% ADE20K mIoU with one glimpse and 84.5% ImageNet accuracy.

Yohaï-Eliel BERREBY, Sabrina Du, Audrey Durand, B. S Krishna

Paris Poster Session 2, Wed, Dec 9, 5:00 PM–7:00 PM, Paris Poster Hall · Published 2026 · ▲ 13 on Hugging Face · Code ★ 23