We are the first to formulate and study the problem of unified video protection against both reference- and tuningbased customization. We identify three key temporal challenges and propose a unified video protection method TC-UAP.

This work provides a basis for the ECCV 2026 LifeGenIP Challenge.

We propose the first NeRF approach that self-calibrates the misalignment between unsynchronized RGB and depth frames, learning an implicit time-pose network jointly with the radiance field.

We survey functionally specialized attention heads in LLMs, categorizing them with a human-inspired four-stage framework (knowledge recalling, in-context identification, latent reasoning, expression preparation) and reviewing how such heads are discovered and evaluated.

We propose an open-source autonomous driving simulator built on neural radiance fields (NeRFs). It is instance-aware, modular, and realistic, achieving state-of-the-art photorealism.