Overview
What this challenge is about.
Use the publicly-released SoccerNet or a synthetic 4-view dataset (you can render with Unity or use a provided one). Implement a 2D pose estimator per view (HRNet or YOLOv8-pose), then triangulate to 3D using known camera calibrations. Evaluate joint-position MPJPE (mean per-joint position error, in cm) against ground truth on a held-out 200-frame test set. Deliver pipeline code, evaluation report, and a 3-page memo on what cost + accuracy trade-offs scale to a 20-stadium rollout.
The Brief
What you'll do, and what you'll demonstrate.
Build a multi-view 3D pose pipeline with MPJPE under 8 cm and lay out the cost + accuracy trade-offs for a 20-stadium scale-up.
Earning criteria — what you'll demonstrate
- Apply 2D pose estimation as a building block for 3D reconstruction
- Use camera calibrations + triangulation for 3D joint recovery
- Evaluate pose estimation with MPJPE and visual sanity checks
- Reason about scaling a vision pipeline across many fixed-camera sites
Program Fit
Where this fits in your program.
Sharpens the same skills your degree expects you to demonstrate.
Skills
Skills you'll demonstrate.
Each one shows up on your verified credential.
Careers
Roles this prepares you for.
Real titles. Real skill bridges. Pick the one closest to your trajectory.
Computer Vision Engineer
Multi-view 3D pose pipelines are exactly the work CV engineers ship at sports-analytics, AR, and motion-capture companies.
This challenge sharpens
- pose-estimation
- multi-view-geometry
- 3d-reconstruction
Machine Learning Engineer
End-to-end pipelines with proper evaluation are the MLE habit that turns research code into product code.
This challenge sharpens
- pytorch
- ml-pipelines
- evaluation
Applied AI Scientist
Translating a pose pipeline into a scale-up plan with cost trade-offs is the applied-AI-scientist craft of bridging research and business.
This challenge sharpens
- evaluation
- pose-estimation
- 3d-reconstruction