Label Ai
Scan Safe, Eat Safe: AI-Powered Allergen Detection in 5 Seconds
Created on 13th April 2025
•
Label Ai
Scan Safe, Eat Safe: AI-Powered Allergen Detection in 5 Seconds
The problem Label Ai solves
Food allergies cause 200,000+ ER visits yearly. Hidden ingredients like “casein” (milk) or “arachis oil” (peanuts) put 10% of adults at risk. Existing solutions fail because:
🚫 Manual entry apps (e.g., Spoonful) are error-prone.
🚫 Generic OCR tools (Google Lens) don’t flag allergens.
🚫 Static databases lack real-time label analysis.
Label AI Fixes This:
✅ Instant Scan: Upload/capture label images → AI extracts + decodes ingredients.
✅ Custom Alerts: Flags 1,000+ allergen synonyms from user-defined profiles.
✅ Health Score: Rates products 0-100 using OpenAI analysis.
Challenges I ran into
OCR Accuracy: Blurry/low-light labels caused extraction errors.
Solution: Preprocessed images with OpenCV (grayscale + sharpening).
Allergen Synonyms: Matching terms like “whey” to “milk”.
Solution: Built a custom database with 1,200+ allergen variants.
Offline Use: Farmers/stores often lack internet.
Solution: Optimized Tesseract.js to work offline.
Technologies used