Nourish

On-Device Nutrition Label Scanner

MOBILE APPIn Progress

About the Project

Nourish (formerly Nutrition Scanner) points the iPhone camera at any nutrition label and locks in protein and calorie values in seconds — entirely on-device, with no backend. A live camera pipeline combines Apple Vision document recognition with custom-trained Core ML models and a deterministic multilingual parser; a value only locks after multiple camera frames agree, keeping wrong reads at zero. The result is shown as protein density on a gamified gauge with 7 tiers from "DEATH" to "SUPERFOOD". Privacy-first by construction: no accounts, no analytics, no network requests — photos never leave the phone.

Key Features

  • Fully on-device label reading: Vision + custom Core ML models

  • Live Scan locks values in seconds via multi-frame agreement voting

  • Multilingual parser (20+ languages, split tables, curved jars)

  • Gamified protein-density gauge with 7 tiers

  • Zero-configuration, no account required

  • No network requests — photos never leave the phone

Tech Stack

Frontend

SwiftSwiftUIAVFoundation

Tools

Apple VisionCore MLStoreKit 2

Tags

#iOS#Swift#Core ML#Vision#Nutrition#SwiftUI