/kada

Primary LanguageTypeScript

K-Pop Formation Generator

Setup

Frontend:

  1. cd frontend
  2. pnpm install
  3. npx next dev

The frontend now runs on https://localhost:3000.

Backend:

  1. cd backend
  2. pip install -r requirements.txt
  3. uvicorn server:app

Features

  • Automatically detects position and translates to a matrix
  • Detects depth
  • Color-codes center
  • 3D view (ThreeJS)
  • Sync frames/poses with music
  • Store user data in Supabase
  • Pose estimation for showing move sequences
  • Service-based arch

Tech Stack

  • Frontend
    • NextJS
    • ThreeJS
    • React
    • Shadcn
  • Backend
    • FastAPI
    • Pytorch
    • GPT-4o
    • Deepface
    • OpenCV
    • MediaPipe (pose estimation)
    • Numpy (position translation)
    • Supabase

Timeline

  • Make Figma (Char)
  • Three.js rendering engine (William)
  • Build UI (William)
  • Facial recognition + persistence (Char & Shresht)
  • Pose estimation + depth (Char & Shresht)
  • Create position (Char & Shresht)
  • API (Char & Shresht)
  • Supabase schemas + storage (Char & Shresht)
  • Connect API to frontend
  • Color coding (gpt 4o)