My research interests lie at the intersection of vision and graphics. Recently, I have been focusing on 3D human motion modeling and generation. Some papers are highlighted.
This paper introduce a novel Recursive Fusion model, dubbed ReFu, designed to integrate point clouds and meshes for exemplar-free 3D Class-Incremental Learning.
This paper introduces SMooDi, which enables stylized motion generation by incorporating style motion sequences into a text-conditioned human motion generation model.
This paper introduces OmniControl, which incorporates flexible spatial control signals into a text-conditioned human motion generation model based on the diffusion process.
This paper proposes an aesthetic-guided outward image cropping method that extends beyond the image border to achieve compositions unattainable with previous methods.
This paper presents a graph-structured representation with BRNN for static-based facial expression recognition, achieving improved accuracy in recognizing facial expressions.