CV Gluon App

Computer vision app with Gluon (MXNet) backend via FastAPI, React frontend (default), and optional Vue frontend.

Setup

1. Backend:

   • cd backend
   • pip install -r requirements.txt
   • uvicorn main:app --reload # Run on http://localhost:8000

2. Frontend (React - Default):

   • cd frontend-react
   • npm install
   • npm start # Runs on http://localhost:3000, proxies to backend

   To switch to Vue:

   • Comment out React serve in scripts.
   • cd frontend-vue
   • npm install
   • npm run serve # Runs on http://localhost:8080

3. Training:

   • POST to /train with JSON {"from_scratch": true/false}
   • Model saves to backend/saved_models/model.params

4. Inference:

   • POST to /predict with form-data file="image"

5. Notes:

   • Vue is disabled by default; enable by running its serve command instead of React.
   • For production, use nginx to serve frontends and proxy to backend.
   • Git init and commit as needed.