Snap Your Meal

• 

  It all starts with a photo. You can generate recipes from your ingredients, estimate calories for a meal or generate a goal-based meal plan

• 

  Upload a photo of your food cupboard and the app will detect all of your ingredients. You can then tailour your recipes to your preferences

• 

  Recipes include ingredients, cooking times, instructions and nutritional information. The app even generates a photo of how it could look.

• 

  You can save your recipes to your Recipe Book and share them with other cooks in the community.

• 

  Upload a photo of any meal and get an instant calorie estimation with a macro split. In future you will be able to generate a recipe too.

• 

  Upload a photo of your ingredients to create a meal plan.

• 

  You can then select the goals you have in mind for your meal plan, such as losing weight or gaining muscle.

• 

  You can also share information about your weight, height, gender and activity level to get a personalised TDEE estimation with your plan.

• 

  A short while later, you have a complet meal plan, with your TDEE. Each recipe can be clicked to view the full details of how to prepare it.

Inspiration

Snap Your Meal was born from a common moment of frustration: staring at a cupboard full of random ingredients and feeling stuck on what to cook. Add to that the rising cost of living, wasted groceries, and the mental load of meal planning—and I saw an opportunity to solve all of it with a single photo.

I also recognized the inefficiencies of calorie-tracking apps and fitness meal planners, which require manual input and don't account for what users actually have on hand. By combining computer vision and AI-powered text generation, I wanted to create something more intuitive, sustainable, and fun.

What it does

Snap Your Meal helps users:

• Identify ingredients from a photo of their cupboard, fridge, or shelf
  
• Generate personalized recipes based on fitness goals and dietary preferences
  
• Estimate calories and macros from a photo of a cooked meal
  
• Build meal plans that match specific calorie and macro targets
  
• Reduce food waste by using ingredients that might otherwise be thrown away
  
• Join a recipe-sharing community of health-conscious home cooks
  

Everything starts with a single snap.

How I built it

I built Snap Your Meal using:

• Frontend: React with TypeScript and Tailwind CSS
  
• Authentication: Supabase Auth for secure user sign-up and login
  
• Database: Supabase PostgreSQL with Row-Level Security
  
• Image Analysis: Supabase Edge Functions powered by GPT-4 Vision for food and meal recognition
  
• Recipe Generation: Another Supabase Edge Function that prompts GPT-4 to create recipes in markdown, based on user-selected preferences and AI-detected ingredients
  
• Calorie Estimation: Custom prompts to extract total calories, macros, and portion estimates from uploaded meal photos
  
• Image Compression: browser-image-compression library for efficient uploads
  
• State Management: Zustand
  
• Routing: React Router
  
• Image Storage: Supabase Storage
  
• AI Recipe Image Generation: GPT-based image generation for recipe previews
  

Challenges I ran into

• Accuracy of vision models: Ingredient recognition can vary based on lighting, image quality, and packaging. I had to craft very specific prompts to improve precision.
• Prompt engineering: Getting the AI to generate detailed, calorie-matched, markdown-formatted recipes required a lot of trial and error.
• AI cost control: Running GPT-4 with vision input is expensive. I had to compress images on the frontend and implement fallback logic to ensure a smooth experience even under rate limits.
• Balancing UX with constraints: I wanted a simple UX (just take a photo), but still needed users to select preferences, goals, and plan types behind the scenes.

Accomplishments that I'm proud of

• Built a fully working app that recognizes ingredients and generates fitness-aligned recipes in under a minute
  
• Created a smooth UX with zero manual input required to get smart, goal-based suggestions
  
• Successfully integrated AI image + text models into a scalable frontend using Supabase
  
• Helped early users reduce food waste, plan their diets, and track nutrition without spreadsheets or calorie math
  
• Designed a freemium model with clear upgrade incentives and community-driven growth potential

What I learned

• Prompt design is a product feature, not a backend detail. Small wording tweaks make huge differences.
• AI is only as useful as its UX wrapper. Simplicity and clarity in user flow are key.
• It's possible to create a valuable, intuitive experience even when dealing with complex tech like computer vision, macros, and dynamic meal planning.

What's next for Snap Your Meal

• Launching community recipe sharing, where users can post, rate, and remix photo-based recipes. A basic version of this exists currently with a few minor bugs to iron out.
  
• Building an app!, I'd love to hire a dev and get some help building this out as an Android and Ios app. I think I could reach and help a lot more people via appstores.
  
• Expanding into language localization for global markets
  
• Exploring API licensing for gyms, nutrition coaches, and health platforms
  
• Applying for sustainability grants and partnerships with anti-food waste orgs
  

Built With

• auth
• css
• gpt-4
• postgresql
• react
• supabase
• tailwind
• typescript
• zustand