About Nutrivision AI

NutriVision AI 🥗📸 Welcome to the central repository for **NutriVision AI**, an AI-powered Calorie Tracker and Nutrition Engine designed for Indian and global cuisines. The platform combines Computer Vision food detection, relative depth estimation, a 3NF relational PostgreSQL database, an Express API with atomic stored procedures, an on-demand Android companion app for Samsung Health synchronization, and a modern React web dashboard. Live Deployments **Production Web Dashboard**: https://nutrivision-ai-mu.vercel.app **Production Backend API**: https://nutrivision-backend-i9x2.onrender.com **Android Companion APK (Direct Download)**: NutriVisionSync.apk (Latest) Project Structure & Technology Stack **Database**: PostgreSQL 15/16 (via Docker & Neon Cloud) with 16 normalized tables in 3NF, 3 triggers, 2 SQL views, and atomic stored procedure sp_log_meal. **Backend API**: Node.js 22, Express, TypeScript, Prisma ORM, Zod validation, JWT authentication (backend-api/, port 3000). **Multimodal AI Vision & Coach**: Google Gemini Multimodal Vision & Reasoning (gemini-3.5-flash-lite, gemini-3.5-flash, gemini-3.8-flash) with dynamic model resilience cascade. **Computer Vision Microservice**: Python 3.11, FastAPI, Roboflow IndianFoodNet API + MiDaS DPT_Hybrid relative depth estimation (ai-service/, port 8000). **Frontend Web Dashboard**: React 19, TypeScript, Vite 8, Tailwind CSS v4…

Release archive