Adam Sun

Hi! I'm Adam Sun.

I'm a senior and 1st-year master's student at Stanford University studying computer science. I'm interested in artificial intelligence research, particularly 3D computer vision for scene understanding and medical applications. I'm passionate about finding new ways to use AI to benefit humanity.

At Stanford, I'm a student researcher at Stanford Vision and Learning Lab (SVL), where I am fortunate to work with Professors Fei-Fei Li and Ehsan Adeli at the Partnership in AI-Assisted Care, as well as with Professor Leonidas Guibas at the Geometric Computation Group. In my work there, I've published papers to NeurIPS 2024 and ICCV 2023 on rendering humans from object-occluded monocular videos using neural radiance fields. I'm also one of the head TAs/Course Organizers for Stanford's CS21SI (AI for Social Good).

When I'm not in class, you can probably find me beatboxing with my a cappella group, sampling the local food scene, or biking aimlessly around campus.

My resume is here. You can also reach me at adsun@stanford.edu.

Education

Stanford University

Sep 2021 - June 2025 (GPA: 4.02/4.0)

B.S. Computer Science, AI track (Advised by Ron Fedkiw)

Clubs: Stanford O-Tone East Asian A Cappella (Vocal Percussionist), CS + Social Good (AI + Social Good Team Lead)

Relevant Coursework: Vision (CS 231N/231A/348I), NLP (CS 224N/324H), AI Principles (CS 221), AI for Biomedical Imaging (CS 235), Systems I/II (CS 107/111), Algorithms (CS 161), Linear Algebra/Multivariable Calculus (MATH 51, CS 205L), Probability (CS 109), Graphics (CS 148, 348I), Information Theory (ENGR 76)

Planned Coursework: Simulation and AI Agents (CS 222)

Work Experience

A list of the things I've done and places I've been.

Stanford Vision Lab

Sep 2022 - Present

I work with Professors Fei-Fei Li and Ehsan Adeli on 3D vision and multimodal learning research for applications in the ICU.

Wells Fargo

Jun 2022 - Sep 2022

I investigated explainable AI for high-risk applications. I wrote a report assessing the fidelity of explanation methods.

Stanford Geometric Computation Lab

Jun 2024 - Present

I work with Professor Leonidas Guibas on 4D content generation and animation with diffusion models.

Ford Motor Company

Jun 2021 - Aug 2021

I processed vehicle sensor data into trajectories and implemented a model for vehicle trajectory prediction.

Whiterabbit.ai

Jun 2023 - Sep 2023

I designed deep learning models that could diagnose breast cancer from digital breast tomosynthesis images with high accuracy.

University of Michigan

Jun 2020 - Sep 2020

I worked with Prof. David Fouhey on using fully-convolutional neural networks for monocular depth estimation.

Publications

Here's a list of my publications. Stay tuned!

OccFusion: Rendering Occluded Humans with Generative Diffusion Priors
Adam Sun*, Tiange Xiang*, Scott Delp, Ehsan Adeli, and Li Fei-Fei
Conference on Neural Information Processing Systems (NeurIPS), 2024

We introduce OccFusion, an efficient yet accurate method that can reconstruct dynamic humans in severely occluded scenes from monocular video
by combining Gaussian splatting with diffusion priors. We propose a three-stage approach consisting of Initialization, Optimization, and Refinement stages.
[Project Page/Paper]

Rendering Humans from Object-Occluded Monocular Videos
Tiange Xiang, Adam Sun, Jiajun Wu, Ehsan Adeli, and Li Fei-Fei
IEEE/CVF International Conference on Computer Vision (ICCV), 2023

We introduce OccNeRF, a neural rendering method that can render humans in severely occluded scenes.
We propose surface-based rendering by integrating geometry and visibility priors.
[Project Page/Paper]

Wild2Avatar: Rendering Humans Behind Occlusions
Tiange Xiang, Adam Sun, Scott Delp, Kazuki Kozuka, Li Fei-Fei, and Ehsan Adeli
arXiv preprint, 2023

We propose Wild2Avatar, an improvement on OccNeRF that uses a novel occlusion-aware scene parametrization method that can render humans behind occlusions without floaters.
[Project Page/Paper]