We achieve action transfer in heterogeneous scenarios with varying spatial structures or cross-domain subjects.

We propose KV-Edit to address the challenge of background preservation in image editing by preserving the key-value pairs of the background, which effectively handles common semantic editing.

We investigate a new problem called narrative action evaluation (NAE) and propose a prompt-guided multimodal interaction framework.

LOGO is a new multi-person long-form video dataset for action quality assessment.

We propose a dynamic Gaussian Splatting method named ManiGaussian for multi-task robotic manipulation, which mines scene dynamics via future scene reconstruction.