Yongsheng Gao

Achieving accurate video object segmentation without manual labels, this framework outperforms traditional methods by maintaining both spatial and temporal coherence.

Local-GS achieves a staggering 7.76x speedup in 3D Gaussian Splatting without sacrificing quality, revolutionizing real-time novel view synthesis.

Traditional frame-wise evaluation fails to capture the nuances of speech-driven facial motion, but a new sequence-alignment approach reveals clearer trade-offs in generative model performance.

Stop identity drift in your talking-head generators without retraining: a simple feedback loop between the generator and encoder stabilizes identity and motion.

Achieve state-of-the-art 6D object pose estimation with 43x speedup by combining mask-aware pose proposals with amodal-driven refinement.