Ao Gao (้ซ˜ๅ‚ฒ)

I'm a first-year Ph.D. student in Computer Applied Technology at East China Normal University (ECNU), advised by Prof. Yuan Xie in Digital Media & Computer Vision Laboratory (DMCV). I also receive supervision from Dr. Jingyu Gong. I graduated with my Master's degree in Computer Science from Donghua University (DHU).

My research focuses on 3D Reconstruction & Editing, Feed-Forward 3D Reconstruction and Neural Rendering.

Email  /  Scholar  /  Github

profile photo

News

2026.04: ๐ŸŽ‰ Our paper GSCompleter has been released.

2025.12: ๐ŸŽ‰ Our project VirtualOccMaker has been open-sourced! Feel free to use it.

2025.09: ๐ŸŽ‰ I've started pursuing my Ph.D. at DMCV-ECNU lab, hoping everything goes well.

2025.07: ๐ŸŽ‰ Our project EmboSceneExplorer has been open-sourced! Feel free to try it.

2025.03: ๐ŸŽ‰ Our paper EasySplat is accepted by ICME2025!

Publications

GSCompleter: A Distillation-Free Plugin for Metric-Aware 3D Gaussian Splatting Completion in Seconds
Ao Gao, Jingyu Gongโ€ , Xin Tan, Zhizhong Zhang, Lizhuang Ma, Yuan Xieโ€ก
arXiv, 2026
project page / arXiv / code(coming soon)

A versatile plugin for 3DGS scene completion, which adopts a novel "Generate-then-Register" paradigm without distillation, achieving robust completion in just a few seconds.

EasySplat: View-Adaptive Learning makes 3D Gaussian Splatting Easy
Ao Gao*, Luosong Guo*, Tao Chen,Zhao Wang,Ying Tai,Jian Yang,Zhenyu Zhang
ICME, 2025
arXiv / code

A novel KNN-based Gaussian adaptive splitting strategy proposed to effectively improve Gaussian reconstruction quality.

Open Source Projects

EmboSceneExplorer: Embodied Scene Explorer for Multimodal Perception and Navigation
Code Contributors: Ao Gao, Luosong Guo, Chaoyang Li, Jiangming Shi, Zilong Xie
Supervisors: Jingyu Gong, Xin Tan, Zhizhong Zhang, Yuan Xieโ€ 
project page / code

A comprehensive multimodal embodied AI system based on Habitat for 3D perception, reconstruction, visual grounding, and navigation in virtual 3D scenes like ScanNet.

VirtualOccMaker: A Blender-based Tool for Automated Occupancy Label Generation in Dynamic Virtual Environments
Ao Gao, Haoming Chen
code

A Blender-based pipeline designed for constructing dynamic virtual scene data, which seamlessly integrates with LabelMaker to automate the generation of semantic occupancy labels.


This homepage's source code is from Jon Barron's website.