Yisheng He (何益升)

Yisheng He is a researcher at Alibaba. He obtained his Ph.D. at HKUST, advised by Prof. Qifeng Chen, Prof. Long Quan, and Dr. Jian Sun.

Email  /  Google Scholar  /  GitHub

🔥 We are now actively hiring research interns. The successful candidates will conduct research to publish at leading international conferences. To apply, please email your CV to: ethanheysh@gmail.com.

I'm interested in 3D Computer Vision, AIGC, Embodied AI, and Digital Avatar.

Forge4D is the first feed-forward model for 4D human Gaussian reconstruction in real world metric scale, and enables novel-view and novel-time synthesis from uncalibrated sparse-view videos in an efficient streaming manner.

PanoLAM is a large avatar model for Gaussian full-head reconstruction from a single unposed image. It utilize coarse-to-fine and dual-branch frameworks that creates Gaussian full-head within a second.

CoProSketch provides prominent controllability and details for sketch generation with diffusion models.

LAM creates animatable Gaussian heads with one-shot images in a single forward pass, which can be reenacted and rendered on various platforms (including mobile phones) in real time.

LaMP is a language-motion pretraining model that advances text-to-motion generation, motion-text retrieval, and motion captioning through aligned language-motion representation learning.

We build a bidirectional control flow between the style and the content for stylized motion generation and enable multimodal style control including text, image, and style motions.

We introduce a hybrid framework that leverages 3D Gaussian representation to advance physical property identification.

We introduce a 2D joint VQVAE to quantize each joint instead of all joints into tokens. A spatial-temporal modeling framework with temporal-spatial 2D masking and 2D attention is also proposed for motion generation.

We enable high-fidelity, transferable, and intensity control for neural field editing.

Our method synthesizes consistent 3D content with fine-grained sketch control.

We introduce a new problem: open-vocabulary 9D object pose and size estimation, a new dataset: OO3D-9D, and a new framework based on vision foundation model to tackle this problem.

A self-supervised framework for category-level object pose and size estimation via differentiable shape deformation, registration, and rendering.

A new open-set few-shot 6D object pose estimation problem: estimating the 6D pose of an unknown object by a few support views without CAD models and extra training. A large-scale synthesis dataset for pre-training and benchmarks for future research.

A generic full flow bidirectional fusion framework for RGBD representation learning, applied to joint instance semantic segmentation and 3D keypoint-based 6D pose estimation.

A brand new dataset to promote the study of instance segmentation for objects with irregular shapes and an affinity-based algorithm to tackle it.

The first deep learning 3D keypoint-based 6D pose estimation algorithm and an overall framework for joint instance semantic segmantation and 3D keypoint detection.

Last updated: October, 2025.

Thanks Dr. Jon Barron for sharing the template code.