Yueqi Duan

HarnessEval-W transforms world model evaluation from mere scoring to a transparent reasoning process that mirrors human judgment.

Tailoring deepfake detection to individual facial characteristics boosts accuracy and adaptability beyond traditional fixed architectures.

UniqueSplat adapts Gaussian representations to specific viewpoints, achieving superior 3D reconstruction and generalization across datasets.

Flow Splatting achieves superior image quality and faster rendering speeds by efficiently modeling dynamic scenes with 4D Gaussian representations.

STORM recovers up to 63.3% accuracy in visual state space models by enforcing spatial awareness in token reduction, transforming how we approach model efficiency.

TivTok redefines video tokenization by enabling the reuse of persistent information across frames, achieving unprecedented compression efficiency with minimal token usage.

RhymeFlow accelerates video generation by allowing non-keyframes to skip denoising steps, achieving faster inference without compromising visual quality.

Existing video world models struggle with long-term memory retention, and MBench exposes their critical limitations while providing a structured path for future improvements.

Forget fixed agent slots and quadratic attention: Gamma-World uses simplex embeddings and sparse hubs to generate interactive multi-agent environments with better fidelity and control, even generalizing from 2 to 4 players without retraining.

Ditch the slow per-scene optimization: SurfelSplat reconstructs surfaces from sparse views in under a second, matching state-of-the-art accuracy with a 100x speedup.

Get simulation-ready assets for robotics and graphics in under a second, without any manual annotation, using a new feedforward approach that jointly learns physical attributes and 3D Gaussian Splatting reconstruction from a single video.