AI Dancing Coach
Gemini-powered dance coaching system with RAG-based technique feedback.
Role: Developer · Hackathon: Gemini 3 Global Hackathon · Timeline: Jan 2026 – Mar 2026
An AI-powered dance coaching system that evaluates a user’s performance by comparing their dance video against reference choreography, then generates technique feedback grounded in dance pedagogy.
Architecture
- Motion comparison module — analyzes the user’s dance video alongside the reference clip and surfaces pose-level differences.
- Gemini integration — interprets motion deltas and turns them into natural-language technique feedback.
- RAG knowledge layer — a Retrieval-Augmented Generation framework that retrieves dance technique knowledge and choreography guidelines, so coaching suggestions are context-aware instead of generic.
Why it matters
Most fitness/dance apps give pass/fail scores. This system gives the kind of qualitative, technique-grounded notes a human coach would — “your weight is staying on the back foot in the turn,” not just “75% match”.