Inspiration

We were tired of the endless cycle of indecision when it came to choosing what to eat. It wasn’t just inconvenient; it often led to poor food choices. This frustration inspired us to create a solution that would simplify meal selection and bring a bit of excitement back to the process.

What it does

Personalized Recommendations: Provides 9 food items tailored to the user’s tastes, using a TF-IDF matrix and item-based collaborative filtering based on the user’s favorite meals. Meal Database Search: Allows users to search through a large dataset of 13,000 meals with advanced filters for allergens and cuisines. Go Crazy Feature: Generates a unique meal using randomly selected ingredients, complete with AI-generated image, title, ingredients list, and cooking instructions using Gemini and Hugging Face APIs. Random Food Generator: Helps users discover new meals by selecting ingredients or letting the app choose a meal completely at random. Favorites Management: Users can add meals to their favorites, view them in a dedicated tab, and remove them as desired. Meal Information Display: Displays detailed meal information, including an image, title, ingredients, and cooking instructions.

How innovate/creative

This project is a game-changer in the world of food and recipe discovery! By harnessing cutting-edge data science techniques like cosine similarity, it offers highly personalized meal recommendations that are tailored to each user’s tastes and preferences. The ability to precompute these recommendations ensures lightning-fast responses, making the app both innovative and efficient. With fun and interactive features like the "craziness" level for meal generation, users can explore new culinary adventures in a way that's both exciting and unique. This isn't just another recipe app—it's a smart, scalable, and dynamic platform that transforms how we think about food!

How we built it

We started by implementing the simplest feature—querying the dataset for meals. From there, we gradually expanded to include more complex features like the recommendation system, filters, random meals, Go Crazy feature (which uses Gemini AI), favorites management, and user authentication. Each new feature built upon the previous work and inspired more ideas, allowing us to create a cohesive and functional application.

Challenges we ran into

First-time Streamlit users: We had to dive deep into the documentation to fully explore Streamlit’s potential. Debugging: Some bugs took longer to fix due to our novice-level software testing skills. AI Limitations: We faced limitations with AI APIs, which hindered our ability to implement features like nutritional value analysis and staggered the process in implementing the go crazy feature. Time Constraints: We struggled to find time to implement more features and improve usability, all while juggling other assignments. Balancing Responsibilities: Managing this project alongside academic commitments was challenging.

Accomplishments that we're proud of

Successfully Integrated AI APIs: We managed to seamlessly integrate Gemini and Hugging Face APIs to generate creative and realistic meal images, titles, and cooking instructions based on random ingredients. Robust Recommendation System: We developed and deployed an effective recommendation system using TF-IDF and item-based collaborative filtering, providing users with highly relevant meal suggestions. Comprehensive Search Functionality: We implemented advanced search capabilities that allow users to find meals by title or ingredients and filter by allergens and cuisines, making the app highly user-friendly. User-Centric Design: We built an intuitive interface with features like the "Go Crazy" meal generator and a random food generator, catering to users who are indecisive or looking for meal inspiration. End-to-End Web Application: Despite time constraints, we built and deployed a full-stack web application with multiple functionalities, including a favorites management system, which we are proud to use ourselves. Scalable Database Management: We successfully set up and managed a MongoDB Atlas Cloud Database to store user information, favorite meals, and facilitate efficient CRUD operations.

Our team

Our team of developers/data scientists worked towards different areas of the project. We had backend developers who ensured that every meal suggestion is as accurate and personalised as possible. This also included handling all the complex integrations like cosine similarity and ensuring that the app runs smoothly behind the scenes. Our front-end developer is the creative force, designing an intuitive and visually appealing interface that makes the user experience seamless and enjoyable. Finally, the role of media generation for our pitch was conducted, ensuring that the best display of our website is optimally shown in its best light.

What we learned

AI Integration: How to implement AI APIs into web apps and link the database to the server. Improved Coding Practices: This project helped us refine our coding skills and adopt better development practices. Streamlit Mastery: We became proficient in using Streamlit for web app development.

What's next for Nyom Nyom

We have big plans for Nyom Nyom! We hope to further enhance the app with more insane and crazy features that, unfortunately, we didn’t have enough time to implement. Our goal is to make Nyom Nyom even more fun, engaging, and useful for users.