|
Email / CV / Linkedin / Github / Scholar I am a second-year PhD student at the CVMI Lab, supervised by Xiaojuan Qi. My research focuses on 3D/4D reconstruction, interaction, simulation, and editing. Prior to this, I completed my master’s degree at the Institute of Computing Technology, part of the Chinese Academy of Sciences, under the supervision of Professor Lin Gao. I deeply appreciate the valuable guidance and support provided by Prof. Gao during my studies. I have also had the privilege of collaborating closely with Dr. Yan-Pei Cao and Professor Yu-Kun Lai, both of whom significantly contributed to my academic growth. Before embarking on my master’s program, I earned my bachelor's degree from the University of Chinese Academy of Sciences, where I was mentored by the esteemed Professor Xilin Chen. Prof. Chen introduced me to my research field and taught me the foundational principles of conducting research, for which I am sincerely grateful.
|
|
|
|
[11] Yihua Huang , MingXian Lin, Yang-Tian Sun, Ziyi Yang, Xiaoyang Lyu, Yan-Pei Cao#, Xiaojuan Qi# IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2025 paper / project page / code We introduce Deformable Radial Kernel (DRK), which extends Gaussian splatting into a more general and flexible framework. Through learnable radial bases with adjustable angles and scales, DRK efficiently models diverse shape primitives while enabling precise control over edge sharpness and boundary curvature |
|
[10] Yihua Huang *, Yang-Tian Sun*, Ziyi Yang*, Xiaoyang Lyu, Yan-Pei Cao#, Xiaojuan Qi# IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2024 paper / project page / code We introduce sparse-controlled gaussian splatting to synthesize dynamic novel views. With the learned node graph of sparse control points, real-time editing can be achieved with ARAP deformation by interactive dragging of users. |
|
[9] Yang-Tian Sun*, Yihua Huang *, Lin Ma, Xiaoyang Lyu, Yan-Pei Cao, Xiaojuan Qi# Conference on Neural Information Processing Systems (NeurIPS), 2024 paper / project page / code We introduce a novel explicit 3D representation—video Gaussian representation—that embeds a video into 3D Gaussians, enabling tracking, consistent video depth and feature refinement, motion and appearance editing, and stereoscopic video generation. |
|
[8] Ziyi Yang, Xinyu Gao, Yang-Tian Sun, Yihua Huang , Xiaoyang Lyu, Wen Zhou, Shaohui Jiao, Xiaojuan Qi# Xiaogang Jin# Conference on Neural Information Processing Systems (NeurIPS), 2024 paper / project page / code We introduce Spec-Gaussian, an approach that utilizes an anisotropic spherical Gaussian (ASG) appearance field instead of SH for modeling the view-dependent appearance of each 3D Gaussian. |
|
[7] Xiaoyang Lyu, Yang-Tian Sun, Yihua Huang , Xiuzhe Wu, Ziyi Yang, Yilun Chen, Jiangmiao Pang, Xiaojuan Qi# ACM SIGGRAPH 2024 Conference Proceedings (SIGGRAPH), 2024 paper We introduce a differentiable SDF-to-opacity transformation function that converts SDF values into corresponding Gaussians' opacities. This function connects the SDF and 3D Gaussians, allowing for unified optimization and enforcing surface constraints on the 3D Gaussians. |
|
[6] Yihua Huang, Yan-Pei Cao, Yu-Kun Lai, Ying Shan, Lin Gao IEEE Transactions on Pattern Analysis and Machine Intelligence (IEEE TPAMI), 2024 paper / project page / code We propose an algorithm to synthesize NeRF textures on arbitrary manifolds. By using a patch-matching method on curved surfaces, we can smoothly quilt texture patches on mesh surfaces. We create a multi-resolution pyramid for a fast patch-matching process. By incorporating a reflection network, we preserve high-frequency view-dependent features such as highlights and mirror reflections in the final synthesized results. |
|
[5] Yihua Huang, Yan-Pei Cao, Yu-Kun Lai, Ying Shan, Lin Gao ACM SIGGRAPH 2023 Conference Proceedings (SIGGRAPH), 2023 paper / project page / code We introduce a NeRF-based system to acquire, synthesize, map, and relight textures from real-world textures. A novel coarse-fine disentangling representation is proposed to model meso-structures of textures. Acquired textures are synthesized by an implicit patch-matching algorithm. |
|
[4] Yihua Huang, Yue He, Yu-Jie Yuan, Yu-Kun Lai, Lin Gao IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2022 arxiv / project page / code We propose a novel mutual learning framework for 3D scene stylization that combines a 2D image stylization network and NeRF to fuse the stylization ability of 2D stylization network with the 3D consistency of NeRF. |
|
[3] Yuting He, Yiqiang Chen, XiaoDong Yang, Hanchao Yu, Yihua Huang, Yang Gu IEEE Transactions on Big Data (TBD), 2022 paper We propose a Selective Self-Distillation method for Federated learning (FedSSD), which imposes adaptive constraints on the local updates by self-distilling the global model's knowledge and selectively weighting it by evaluating the credibility at both the class and sample level. |
|
[2] Yu-Jie Yuan, Yu-Kun Lai, Yihua Huang, Leif Kobbelt, Lin Gao, IEEE Transactions on Pattern Analysis and Machine Intelligence (IEEE TPAMI), 2022 paper / project page We introduce a novel NeRF reconstruction method using RGB-D inputs from a consumer-level device (iPad), which enables high-quality reconstruction from sparse inputs. Experiments show that the proposed method achieves state-of-the-art novel view synthesis quality in this case of sparse RGB-D inputs. |
|
[1] Jie Yang, Lin Gao, Qingyang Tan, Yihua Huang, Shihong Xia Yu-Kun Lai IEEE Transactions on Visualization and Computer Graphics (TVCG), 2021 arxiv We propose a novel method to exact multiscale deformation components automatically with a stacked attention-based autoencoder. |
|
Internships: Tencent (2023 Summer) |
|
Kudos to Dr. Jon Barron for sharing his website template. |