FoodVision Pro

AI nutrition coach in your pocket: scan any meal, get instant nutrition data, track health trends, and prevent diabetes with smart insights

Comment

Inspiration

I am a Diabetes and Endocrinology Consultant and deal with metabolic issues daily.

  • Many patients struggle to track their food accurately, not because they lack motivation, but because the process is time-consuming and confusing.
  • Portion sizes are often guessed wrong, leading to poor decisions despite good intentions.
  • Most nutrition apps are abandoned within weeks due to friction and overwhelm.
  • One moment stood out: a patient said, "I’ve tried every app. I just can’t keep up."
  • That made us realize, it’s not about effort, it’s about making the experience simpler.

What it does

  • Lets users take a photo of their meal and instantly view nutrition details.
  • Uses AI to recognize food with high accuracy (85–92% for common dishes).
  • Provides real-time diabetes risk estimates based on the meal and user profile.
  • Automatically tracks macronutrients and visualizes trends over time.
  • Designed to reduce food logging time from minutes to just a few seconds.

How we built it

  • Started with real patient feedback about daily tracking challenges.
  • Used React Native for cross-platform mobile development (via Lovable.dev).
  • Integrated OpenAI’s GPT-4 Vision API for food recognition.
  • Built backend services using Supabase for authentication and data storage.
  • Plan to integrate RevenueCat for managing in-app subscriptions across app stores.

Challenges we ran into

  • Food recognition is challenging, especially with mixed or homemade meals.
  • Balancing AI accuracy and a simple user experience took careful iteration.
  • App store payment rules were more complex than expected, still a work in progress.
  • Maintaining clinical value without overwhelming users required thoughtful design.
  • Scaling OpenAI usage cost-effectively was a key technical concern.

Accomplishments that we're proud of

  • Developed a working MVP that delivers real-time photo-to-nutrition analysis.
  • Achieved fast response times of 2–3 seconds per food scan.
  • Reduced meal logging time by over 95% compared to traditional methods.
  • Incorporated ADA-aligned diabetes risk scoring to ensure clinical relevance.
  • Built something that people genuinely enjoy using, not just another tracking app.

What we learned

  • Simplicity beats feature-packed designs in health-focused apps.
  • Visual, instant feedback is much more engaging than traditional logs.
  • Even small barriers can cause users to give up on long-term habits.
  • Confidence scoring helps build trust in AI-powered decisions.
  • You can blend AI with medical insights in a way that feels approachable, not overwhelming. ## What's next for FoodVision Pro
  • Launching the free version on iOS and Android to gather real-world feedback.
  • Running pilot programs to validate usability in clinical and everyday settings.
  • Education & Prevention: Introduce guided programs focused on diabetes prevention and to help users build lasting habits.
  • Enhance food recognition accuracy with feedback loops and ongoing training.
  • Explore features like meal planning, smart grocery suggestions, and food label decoding.
  • Long-term vision: global expansion and integration with healthcare systems and EMRs.

Built With

Share this project:

Updates