GKDT: General Keypoint Detection Transformer

Changsheng Lu, Yuxin Chen, Haokun GUI, Rong Wang, Jie Yang, Harry Yang, Anton van den Hengel, Jiaya Jia

• We present MegaKPT, a large-scale unified keypoint dataset, and GKDT, a flexible DINOv3-based Transformer for general keypoint detection, supporting visual and text prompts to detect keypoints across broad seen and unseen object categories.

Towards Realistic and Consistent Orbital Video Generation via 3D Foundation Priors

Rong Wang, Ruyi Zha, Ziang Cheng, Jiayu Yang, Pulak Purkait, Hongdong Li

Video

• We present a novel method for generating geometrically realistic and consistent orbital videos from a single image, leveraging rich shape priors from a 3D foundation model to improve structural coherence, multi-view consistency, and generalization to complex camera trajectories.

Learning High-Fidelity Garment Deformation via Skinning-Free Image Transfer

Rong Wang, Wei Mao, Changsheng Lu, Hongdong Li

• We present a skinning-free image transfer framework for high-fidelity 3D garment deformation, decoupling low-frequency posed shapes and high-frequency wrinkle details to generate realistic animation across garments with diverse topologies.

FRESA: Feedforward Reconstruction of Personalized Skinned Avatars from Few Images

Rong Wang, Fabian Prada, Ziyan Wang, Zhongshi Jiang, Chengxiang Yin, Junxuan Li, Shunsuke Saito, Igor Santesteban, Javier Romero, Rohan Joshi, Hongdong Li, Jason Saragih, Yaser Sheikh

GitHub, Project

• We learn a universal prior from over a thousand clothed humans to achieve instant feedforward generation and zero-shot generalization, aiming for reconstructing personalized 3D human avatars with realistic animation from only a few images.

Towards High-Quality 3D Motion Transfer with Realistic Apparel Animation

Rong Wang, Wei Mao, Changsheng Lu, Hongdong Li

GitHub

• We present a novel method aiming for high-quality motion transfer with realistic apparel animation, and also build a new dataset named MMDMC combining stylized characters from the MikuMikuDance community with real-world Motion Capture data.

DeepSimHO: Stable Pose Estimation for Hand-Object Interaction via Physics Simulation

Rong Wang, Wei Mao, Hongdong Li

GitHub

• We present DeepSimHO: a novel deep-learning pipeline that combines forward physics simulation and backward gradient approximation with a neural network.

Interacting hand-object pose estimation via dense mutual attention

Rong Wang, Wei Mao, Hongdong Li

GitHub

• We propose a novel dense mutual attention mechanism that is able to model fine-grained dependencies between the hand and the object.

Transcribing natural languages for the deaf via neural editing programs

Dongxu Li, Chenchen Xu, Liu Liu, Yiran Zhong, Rong Wang, Lars Petersson, Hongdong Li

• We design a new neural agent that learns to synthesize and execute editing programs, conditioned on sentence contexts and partial editing results.

AttentionFlow: Visualising Influence in Networks of Time Series

Minjeong Shin, Alasdair Tran, Siqi Wu, Alexander Mathews, Rong Wang, Georgiana Lyall, Lexing Xie

GitHub

• We present AttentionFlow, a new system to visualise networks of time series and the dynamic influence they have on one another.