Haolin Xiong

I'm a PhD student in Computer Science at the University of Southern California - Institute for Creative Technologies (USC-ICT), advised by Assistant Professor Yajie Zhao.

Previously, I earned my M.S. in Electrical and Computer Engineering from UCLA, where I was advised by Professor Achuta Kadambi and Professor Lin Yang. I also hold an M.S. in Business Analytics and dual B.S. degrees in Computer & Systems Engineering and Economics from Rensselaer Polytechnic Institute.

CV  /  Linkedin  /  Twitter  /  Github

Portrait of Haolin Xiong

Research

My research interests broadly span 3D vision and perception, with occasional work on 2D vision models such as generative foundation models. Recently, I have been focusing on 3D scene understanding using 3DGS, along with its potential applications in robotics and VR.

Mind-to-Face: Neural-Driven Photorealistic Avatar Synthesis via EEG Decoding


Haolin Xiong, Tianwen Fu, Pratusha Prasad Yunxuan Cai, Haiwei Chen, Wenbin Teng, Hanyuan Xiao, Yajie Zhao

2026 (Under Review)
Reconstruct 3D head avatar with expression from EEG signals.

Olbedo: An Albedo and Shading Aerial Dataset for Large-Scale Outdoor Environments


Shuang Song, Debao Huang Deyan Deng, Haolin Xiong, Yang Tang, Yajie Zhao, Rongjun Qin

CVPR, 2026
The first large-scale, real-world albedo dataset designed for weather-agnostic 3D asset creation from aerial imagery.

SparseGS: Sparse View Synthesis with 3D Gaussian Splatting


Haolin Xiong, Sairisheek Muttukuru, Hanyuan Xiao, Rishi Upadhyay, Pradyumna Chari, Yajie Zhao, Achuta Kadambi
project page / arXiv

3DV, 2025
Real-world scene reconstruction using 3D Gaussian Splatting with 12 input views.

Localized Gaussian Splatting Editing with Contextual Awareness


Hanyuan Xiao, Yingshu Chen, Huajian Huang, Haolin Xiong, Jing Yang, Pratusha Prasad, Yajie Zhao
arXiv

WACV, 2025
Contextual and lighting-aware 3D editing with Gaussian Splatting.