Overview
What this challenge is about.
Build a Python pipeline: face detection + landmark extraction, FLAME (or equivalent) face-model fitting to landmarks + photometric loss, texture inference, and skinning weight assignment. Validate on the 30-person test cohort (provided, stratified across 4 skin tones, 2 genders, 4 age bands). Report per-stratum geometric reconstruction error and a human-evaluated likeness score from a 5-rater panel. Measure end-to-end latency on a single A100-class GPU. Deliver source, the 30 generated avatars, a fairness + quality report, and a 10-page design document explaining each stage.
The Brief
What you'll do, and what you'll demonstrate.
Build a single-selfie-to-riggable-avatar pipeline that runs under 90 seconds, validated fairly across a 30-person stratified test cohort.
Earning criteria — what you'll demonstrate
- Compose a digital-human pipeline from publicly-available components
- Fit a parametric face model to a single image
- Evaluate avatar quality fairly across demographic strata
- Communicate fairness findings honestly to a product audience
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.
Product Manager
PMs who understand the fairness realities of avatar pipelines scope features that ship to all users instead of breaking in the first demographic audit.
This challenge sharpens
- fairness-evaluation
- digital-humans
- 3d-rendering