Overview
What this challenge is about.
Use the nuScenes or Waymo Open Dataset (open access) as your training and evaluation source. Fine-tune a strong baseline (e.g., CenterPoint or BEVFusion) and define an evaluation slice for 'parked vehicle within 3 m of the road edge'. Report mean Average Precision (mAP) overall and on the shoulder-vehicle slice, plus per-class regression checks. Deliver a model checkpoint, an evaluation report, and a 4-page memo for perception leadership.
The Brief
What you'll do, and what you'll demonstrate.
Fine-tune a 3D detector so the shoulder-vehicle slice mAP improves by at least 5 points without regressing other categories.
Earning criteria — what you'll demonstrate
- Fine-tune a state-of-the-art 3D detector on a public AV dataset
- Define and evaluate operationally-relevant evaluation slices
- Diagnose regression risk across categories
- Communicate perception trade-offs to engineering leadership
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
Fine-tuning 3D detectors with slice-aware evaluation is the day-one perception work at any AV company.
This challenge sharpens
- 3d-object-detection
- perception
- transfer-learning
Machine Learning Engineer
Reproducible training pipelines and rigorous regression checks are the MLE habits that get safety-critical models into production.
This challenge sharpens
- ml-pipelines
- pytorch
- slice-evaluation
AI Safety Researcher
Operationally-defined evaluation slices for failure modes is the safety discipline AV safety researchers practice every quarter.
This challenge sharpens
- slice-evaluation
- perception
- 3d-object-detection