How junioraguilar/oxiom works
Project type and purpose
• This is a full-stack web application for training, testing, and uploading YOLOv8 object-detection models.
• The backend is a Flask API that orchestrates model training/inference, and the frontend is a React (Vite) single-page application that provides the UI.
File structure
• Top-level: Docker-related files (Dockerfile, docker-compose.yml, nginx.conf) and environment definitions (environment.yml).
• backend/: Flask application (app.py), model definitions (models.py), training utilities, local SQLite DB (models_db.sqlite), and Python dependency list (requirements.txt).
• frontend/: React + TypeScript code under src/, HTML entrypoint, configuration (vite.config.ts, tsconfig.json), UI components, pages, API wrapper (src/api/axios.ts), and theming.
• Ignored (in README): models/, uploads/, and train_data/ for persisted model files, user uploads, and raw training data.
README for additional insights
• Describes a Docker-Compose setup spinning up two services: a Flask + YOLOv8 backend (GPU-enabled) and a React dev server.
• Workflow: user uploads or selects a dataset → backend trains YOLO model → backend stores trained model → user tests the model on images → results displayed.