DSL-Lab/MoFlow

[CVPR 2025] MoFlow: One-Step Flow Matching for Human Trajectory Forecasting via Implicit Maximum Likelihood Estimation Distillation

38
/ 100
Emerging

This project helps predict the future movements of people or objects in a scene, like basketball players on a court or pedestrians in a park. You provide past trajectory data, and it generates multiple plausible future paths for all agents, showing both the most likely route and alternative possibilities. It's designed for researchers or analysts working with human behavior modeling and spatial prediction.

107 stars.

Use this if you need to accurately forecast diverse future trajectories for multiple interacting agents, such as in sports analytics, urban planning, or autonomous systems research.

Not ideal if you only need a single, deterministic prediction and don't require understanding the uncertainty or generating a range of possible future movements.

human-trajectory-forecasting sports-analytics pedestrian-dynamics multi-agent-prediction movement-modeling
No Package No Dependents
Maintenance 6 / 25
Adoption 9 / 25
Maturity 16 / 25
Community 7 / 25

How are scores calculated?

Stars

107

Forks

5

Language

Python

License

MIT

Last pushed

Dec 15, 2025

Commits (30d)

0

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/diffusion/DSL-Lab/MoFlow"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.