Wanning He

I build generative models of human motion, with sports as my testbed.

I'm a master's student in Computer Science at Dartmouth College, where I work in the Visual AI Lab advised by Prof. Yu-Wing Tai. Before Dartmouth, I studied statistics (M.S., Shanghai University of Finance and Economics) and mathematics (B.S., Northeastern University, China), and spent two years as a quantitative researcher at Fourier Capital, a quantitative hedge fund.

I am applying to PhD programs for Fall 2027.

Email / CV / GitHub / Blog

News

11.2026	Awarded Dartmouth's Class of 1961 Arts Initiative Fund for In Her Corner, a photography project on women in boxing.
10.2026	Won First Prize at two NYC hackathons: VAST Builders Challenge (FightLens) and Daytona HackSprint (Just Standup); featured in Dartmouth CS Kudos: Fall 2026.

Research

My research interests lie in building intelligent machines that can perceive, understand, and generate human motion in the dynamic physical world, with sports as a primary testbed. * indicates equal contribution.

RefRoute: Decoupling Conditioning Cost from References via Compact Residual Conditioning and Spatial Routing
Wanning He*, Yuyao Zhang*, Yu-Wing Tai
arXiv preprint, 2026
arXiv

Compact residual conditioning and spatial routing make multi-reference generation scale with references: 18.3× faster than FLUX.2-Klein-9B at 16 references.

Beyond MoCap: Scaling Motion Tokenizers with Synthetic Human Motion for Generative Modeling
Yiwen Yan, Wanning He, Yu-Wing Tai
arXiv preprint, 2026
demo video / arXiv

Scaling motion tokenizers with synthetic human motion: the bottleneck is the support of the learned representation, not codebook size. The new tokenizer improves T2M-GPT, MotionGPT, and Motion-Agent with no architecture change.

Sports AI Projects

FightLens: Real-Time Combat Sports Analysis
First Prize, VAST Builders Challenge, New York City, 2026
demo / code

A real-time MMA viewing assistant: fighter tracking, pose overlays, received-strike heatmaps, and evidence-based event analysis.

Just Standup: Exercise Anywhere, Anytime
First Prize, Daytona HackSprint w/ Arize AI, New York City, 2026
devpost

No equipment, no place, no coach: only a phone. Boxing with the phone's motion sensors, the city as a gym, and an AI 3D coach.

Awards and Funding

2026	Class of 1961 Arts Initiative Fund ($3,211) for In Her Corner	Dartmouth College
2026	First Prize (team, ~$6,950), VAST Builders Challenge for FightLens	New York City
2026	First Prize (team, $3,000), Daytona HackSprint w/ Arize AI for Just Standup	New York City
2026	Both hackathon wins featured in Kudos: Fall 2026	Dartmouth CS
2026	NSF ACCESS Discover Allocation (PI)	NSF ACCESS
2026	OpenAI Researcher Access Program ($1,000 API credits)	OpenAI
2026	Merit Scholarship (100% tuition)	Dartmouth College
2025	Tuition Scholarship (75% tuition)	Dartmouth College
2021	University-level Outstanding Student (top 3%)	China
2019	Provincial Outstanding Graduate (top 0.2%)	China
2017	National Scholarship (top 0.2%, ~$1,100)	China

Miscellanea

Teaching	Teaching Assistant, Machine Learning and Statistical Analysis (COSC 274), Fall 2026 and Winter 2026 Teaching Assistant, Graph-based Machine Learning (COSC 89.39/189.39), Spring 2026 Teaching Assistant, Deep Learning (COSC 78/278), Fall 2025
Service	Journal Reviewer, International Journal of Computer Vision (IJCV), 2026

Beyond Research

Outside the lab, I'm usually training or outdoors.
Sports: weightlifting, Brazilian jiu-jitsu (Jiangsu Championship silver), boxing, CrossFit, climbing
Outdoors: diving, hiking, camping, travel
Creative: In Her Corner, a photo series on women in boxing · sports vlog (190K+ views)

Website template from Jon Barron.