Biography
[more]
Tiange Xiang
(向天戈)
is a CS Ph.D. student at Stanford University and a visiting student at MIT working with Kaiming He. He is affiliated with Stanford AI Lab & Stanford Vision and Learning Lab. He is
advised by Fei-Fei Li,
co-advised by Scott Delp and Ehsan Adeli.
Previously, he worked with Weidong Cai at The University of Sydney,
where he was awarded the University Medal.
His research focuses on visual intelligence.
He is a recipient of Stanford HAI fellowship and a finalist of the Qualcomm Innovation Fellowship.
Opportunities
I am actively looking for student collaborators to co-lead the 3D Gen Playground project. Please drop me an email if you are a Stanford student and have experience in 3D vision and/or generative models.
Open Source Projects
A user-friendly codebase for accelerating 3D generation research. Features open data platform with standardized protocols, efficient data loaders, interactive 3DGS viewer, and plug-and-play components built on the GaussianVerse dataset.
Publications
[* indicates equal first authorship, † indicates equal last authorship.]
QuantiPhy: A Quantitative Benchmark Evaluating Physical Reasoning Abilities of Vision-Language Models
The first benchmark that tests vision-language models' physical reasoning abilities.
ViBES: A Conversational Agent with Behaviorally-Intelligent 3D Virtual Body
A speech-language-behavior model that jointly plans language and movement for dialogue-conditioned body actions.
Repurposing 2D Diffusion Models for 3D Shape Completion
[ArXiv 2025]
[Paper]
2D diffusion models can also complete 3D point clouds!
SocialGen: Modeling Multi-Human Social Interaction with Language Models
The first unified motion-language model capable of modeling interaction behaviors among varying numbers of individuals.
Repurposing 2D Diffusion Models with Gaussian Atlas for 3D Generation
2D diffusion models are also 3D content generators!
NeuHMR: Neural Rendering-Guided Human Motion Reconstruction
Human mesh recovery guided via generalizable neural rendering.
Wild2Avatar: Rendering Humans Behind Occlusions
Higher-fidelity human rendering from monocular occluded videos!
OccFusion: Rendering Occluded Humans with Generative Diffusion Priors
2D generative priors help inpaint occluded 3D humans.
Exploiting Structural Consistency of Chest
Anatomy for Unsupervised Anomaly Detection
in Radiography Images
A faster and stronger network for chest X-ray anomaly detection.
Rendering Humans from Object-Occluded Monocular Videos
We rendered human from object occluded videos.
SQUID: Deep Feature In-Painting for Unsupervised Anomaly Detection
We re-formulated unsupervised anomaly detection as semantic-sapce in-painting.
DDM2: Self-Supervised Diffusion MRI Denoising with Generative Diffusion Models
We achieved self-supervised MRI denoising through generative diffusion models.
Seeing Beyond the Brain: Conditional Diffusion Model with Sparse Masked Modeling for Vision Decoding
We decoded photo-realistic visual stimuli from fMRI brain signals.
Towards Bi-directional Skip Connections in Encoder-Decoder Architectures and Beyond
An efficient and light-weight encoder-decoder network with SOTA performances.
DSNet: A Weakly-Supervised Dual-Stream Framework for Effective Gigapixel Pathology Image Analysis
We proposed to combine global-local clues for weakly-supervised WSI analysis.
Walk in the Cloud: Learning Curves for Point Clouds Shape Analysis
We proposed a geometry-aware feature aggregation operator for point cloud analysis.
BiX-NAS: Searching Efficient Bi-directional Architectures for Medical Image Segmentation
We proposed a NAS method to search efficient bi-directional architectures.
Partial Graph Reasoning for Neural Network Regularization
[ArXiv 2021]
[Paper]
Two-Stage Monte Carlo Denoising with Adaptive Sampling and Kernel Pool
[ArXiv 2021]
[Paper]
BiO-Net: Learning Recurrent Bidirectional Connections for Encoder-Decoder Architecture
We proposed bi-directional skip connections in the encoder-decoder architecture.
Education
Ph.D. Student in Computer Science
Palo Alto, CA, U.S.
M.S. in Computer Science (petition during Ph.D.)
Palo Alto, CA, U.S.
B.S. in Computer Science and Technology (Advanced) (Honours)
Sydney, NSW, Australia
Service
Organizer:
GenAI4Health Workshop @ NeurIPS 2025
Conference reviewer:
CVPR, ICCV, ECCV, ICLR, ICML, NeurIPS, SIGGRAPH, MICCAI, WACV, AAAI, AISTATS, KDD.
Jounral reviewer:
IEEE TPAMI, IEEE TIP, IEEE TMI, IEEE IV, Neurocomputing, Scientific Reports.
Teaching assistant:
CS 231n (Stanford, Spring 2026 & 2025 & 2024 & 2023), COMP 3419 (USYD, Fall 2019)