Have you ever felt that rush of excitement when you discover a new dish—but then found yourself squinting at an unhelpfully brief menu description, wishing you could see exactly what you’re about to order? Enter PicMenu, an AI-powered tool that transforms restaurant menus into vivid culinary experiences. By simply snapping a photo of a menu, PicMenu generates high-quality images and in-depth descriptions for each dish, bringing the art of dining to life.
Core Features
1. Menu Parsing and Generation
All it takes is a photograph of a paper menu. PicMenu leverages advanced AI to extract each menu item and then creates a detailed, high-quality image that represents the dish. This feature is designed to provide diners with a mouthwatering preview and to minimize the guesswork when ordering.
2. AI-Supported Details
Beyond just pictures, PicMenu can parse and display supplementary information such as potential ingredients, origins, and even calorie counts or flavor profiles (in the near future). Whether you’re counting calories or tracking down that perfect blend of spices, this level of detail can revolutionize how you engage with a menu.
3. Multilingual & Tagging Support (In Development)
If you’ve ever traveled abroad, you know how frustrating it can be to interpret a menu in an unfamiliar language. PicMenu aims to solve this by planning multilingual menu parsing. We’re also developing the ability to tag dishes (e.g., “spicy,” “vegetarian,” “vegan”) for easy filtering, so you can quickly find the meal that best suits your dietary preferences or adventurous tastes.
Technical Architecture
- AI Models
- Llama 3.2 Vision 90B for robust menu information extraction.
- Llama 3.1 8B for generating structured, JSON-formatted outputs.
- Flux Schnell to produce realistic, high-resolution dish imagery.
- Frameworks and Tools
- Built with Next.js and TypeScript, ensuring efficient and scalable web performance.
- Styled using Tailwind CSS and Shadcn, offering a consistent, modern user interface.
- Plausible and Helicone integrations provide analytics and monitoring, so we can continuously refine your user experience.
- File Storage
- AWS S3 hosts all uploaded content and generated assets, ensuring reliability and security.
How to Use
- Clone the Repository
git clone https://github.com/Nutlope/picmenu - Configure Environment Variables
- Create a
.envfile to store your Together AI API key. - Add AWS S3 credentials to enable secure file storage.
- Create a
- Install Dependencies and Run
npm install npm run devThis spins up a local development environment where you can test PicMenu’s features and get a feel for its capabilities.
Future Plans
- More Realistic Imagery: We’re working on enhancing dish visuals so that the AI-generated images look even more mouthwatering and true-to-life.
- Infinite Scrolling & Error Handling: Better user experience through seamless content loading and robust error management.
- Sample Menus in Multiple Languages: Provide example data so you can test PicMenu’s multilingual parsing more extensively.
- Enhanced Filtering: Use dish tags (e.g., “gluten-free,” “kid-friendly,” etc.) to quickly narrow down your menu preferences.
Potential Applications
- Restaurants: Quickly convert paper menus into appealing digital displays, complete with images and detailed descriptions.
- Online Dining Platforms: Offer an immersive browsing experience by showcasing vibrant images of each dish.
- Multilingual Contexts: Ensure global accessibility by translating and tagging menu items for international customers.
At the heart of PicMenu is the belief that dining should be an experience full of delight and exploration. We’re harnessing the power of AI to enhance that special moment when you discover a new dish or revisit a classic favorite. By blending technical rigor with a sprinkle of culinary magic, PicMenu empowers restaurants, online platforms, and food lovers everywhere to embark on a more visual and intuitive gastronomic journey.
For more information, check out the PicMenu Website or visit our GitHub Repository. Bon appétit!
