Description
AI-powered animation is only as good as the data behind it, but most motion data wasn't designed for AI. In this session, we'll demonstrate a projection-mapped capture app that guides performers in real time to produce more-structured motion. As well as how Label Maker, a tool integrated within the Autodesk ecosystem, uses AI to turn raw motion data and videos into richly tagged, reusable data sets. You will walk away with a clearer picture of how to make motion data actually usable for AI, and how to rethink their pipelines to get there. The real value of motion data isn't in the capture it's in the enrichment.
Key Learnings
- Discover gaps between raw motion data and AI-ready data sets, and where value is created or lost in pipelines.
- Learn about principles for systematic data capture to create comprehensive, well-organized motion data sets from the start.
- Learn about and evaluate how AI-assisted annotation reduces manual effort while producing structured data sets AI proposes, human confirms.
- Learn how enriched data sets support AI uses from motion generation to action recognition via “label once, train anywhere.â€