Yuhao Zhang

First-year PhD @imperialcollege

[Yuhao Zhang @yzhanglp]

Hi! I am a third-year Undergraduate student majoring in CS & MATH in HKUST. My research interest lies in CV & ML. I am now doing research guided by Chi-Keung Tang and Yu-Wing Tai.
Feel free to chat with me!
Personal Website: https://t.co/QNTYZdLxe1
#hkust #CVPR2024 #ECCV2024

[Yuhao Zhang @yzhanglp]

More results and all data, code, and models can be found on our project page: https://t.co/R9nWNstAz0

This work will be presented in #SIGGRAPH2025.

This is a joint effort of an amazing team: @yyfdeng, @yzhanglp, @gengchen01, @elliottszwu, @jiajunwu_cs.

[Yuhao Zhang @yzhanglp]

🪄Introducing Anymate—a large-scale dataset of 230K 3D assets with rigging and skinning annotations!

With this dataset, we trained an auto-rigging model and benchmarked a variety of architectures!

🔥Turn static assets into animatable ones in seconds: https://t.co/77ApgJtia1

[Yuhao Zhang @yzhanglp]

We designed a data-driven auto-rigging framework with three sequential modules predicting joints, connectivity, and skinning weights from a sampled point cloud of a given 3D mesh. For each module, we experimented with a variety of architectures and benchmarked their performance.

[Yunzhi Zhang @zhang_yunzhi]

This is a great opportunity to join @elliottszwu's new group! Working with Elliott has always been inspiring and fun---He brings incredible insights and depth to research. Excited to see what the lab will bring!

[Yufan Deng @yyfdeng]

I am a junior Undergraduate student majoring in CS at @hkust, working with Prof. Chi-Keung Tang and Prof. Yu-Wing Tai. I am now doing summer research internship at @Stanford on the topic related to 3D vision under the guidance of @elliottszwu, @gengchen01, and Prof.@jiajunwu_cs

[Yuhao Zhang @yzhanglp]

Our paper "DragVideo: Interactive Drag-style Video Editing" is accepted at #ECCV2024! @eccvconf
The first paper about how to preciously edit video with simple drags and optional masks
Great to work with @yyfdeng and @RickyRDWang
ArXiv: https://t.co/P5nQVa81he

[AK @_akhaliq]

DragVideo: Interactive Drag-style Video Editing

paper page: https://t.co/hBDhmzAEyO

Editing visual content on videos remains a formidable challenge with two main issues: 1) direct and easy user control to produce 2) natural editing results without unsightly distortion and artifacts after changing shape, expression and layout. Inspired by DragGAN, a recent image-based drag-style editing technique, we address above issues by proposing DragVideo, where a similar drag-style user interaction is adopted to edit video content while maintaining temporal consistency. Empowered by recent diffusion models as in DragDiffusion, DragVideo contains the novel Drag-on-Video U-Net (DoVe) editing method, which optimizes diffused video latents generated by video U-Net to achieve the desired control. Specifically, we use Sample-specific LoRA fine-tuning and Mutual Self-Attention control to ensure faithful reconstruction of video from the DoVe method. We also present a series of testing examples for drag-style video editing and conduct extensive experiments across a wide array of challenging editing tasks, such as motion editing, skeleton editing, etc, underscoring DragVideo's versatility and generality.

[Yiping Lu @2prime_PKU]

NEW! https://t.co/hRkOylE9gw Kernel interpolator in fixed dimension can just tempered overfitting, however, we discover that physics-informed learning may make interpolators bengining overfitting! joint work:amazing student @MH2023ML (finding Ph.D) and coauthor @Fanghui_SgrA

[Yuhao Zhang @yzhanglp]

I have started my summer research Internship at @Stanford under the guidance of @elliottszwu , @gengchen01, and Prof.@jiajunwu_cs.
Hope to see you in CA!

[Yuhao Zhang @yzhanglp]

@jeleechandayo JELEE😭

[AK @_akhaliq]

DragVideo: Interactive Drag-style Video Editing

paper page: https://t.co/hBDhmzAEyO

Editing visual content on videos remains a formidable challenge with two main issues: 1) direct and easy user control to produce 2) natural editing results without unsightly distortion and artifacts after changing shape, expression and layout. Inspired by DragGAN, a recent image-based drag-style editing technique, we address above issues by proposing DragVideo, where a similar drag-style user interaction is adopted to edit video content while maintaining temporal consistency. Empowered by recent diffusion models as in DragDiffusion, DragVideo contains the novel Drag-on-Video U-Net (DoVe) editing method, which optimizes diffused video latents generated by video U-Net to achieve the desired control. Specifically, we use Sample-specific LoRA fine-tuning and Mutual Self-Attention control to ensure faithful reconstruction of video from the DoVe method. We also present a series of testing examples for drag-style video editing and conduct extensive experiments across a wide array of challenging editing tasks, such as motion editing, skeleton editing, etc, underscoring DragVideo's versatility and generality.