Inspiration

Cooking and meal planning can be stressful for many reasons. People often watch YouTube videos or follow random online recipes, only to realize they don’t have the right ingredients or that the recipes are too complex or time-consuming. Pantry items are forgotten or go unused, leading to wasted TIME and wasted INGREDIENTS. Many users struggle with deciding what to cook each day, planning balanced meals, or even tracking what’s already in their pantry. These issues create frustration, increase food waste, and make daily meal planning overwhelming. We wanted to solve these problems by making a tool that helps users cook smarter, plan meals efficiently, and make the most of the ingredients they already have. That’s how PantryPilot came to life a vision-powered cooking assistant designed to help people manage their pantry and meals without the stress, save What It Does PantryPilot looks at your pantry using computer vision and figures out what ingredients you have. Then, it suggests recipes you can actually make. You can tell it your dietary preferences, how much time you have, or what type of meal you want, and it gives you personalized meal plans, shopping lists, and ingredient substitutions. Basically, it helps you save food, save money, and cook more confidently. How We Built It We built PantryPilot as a full-stack platform: Frontend: React + Tailwind CSS for a clean and interactive UI. Backend: Flask + Python handles ingredient recognition and recipe suggestions. Machine Learning: A computer vision model detects pantry items and their amounts. A simple recommendation engine matches your ingredients to recipes. A computer vision model detects pantry items and their amounts. A simple recommendation engine matches your ingredients to recipes. Database: Firebase stores pantry data, user preferences, and recipe history. It took a few tries to get the dashboard right, but now it shows real-time recipe suggestions and meal planning summaries clearly. Challenges We Faced One tricky part was teaching the system to recognize ingredients under different lighting. Another challenge was making the recommendations feel useful and not overwhelming. Finally, designing the UI to be simple but informative required some trial and error. Accomplishments During the hackathon, we managed to: Build a working platform that scans pantries and suggests recipes. Make a recommendation engine that adapts to what ingredients are available. Create dashboards showing recipes, shopping lists, and substitutions. Help users reduce food waste, save money, and plan meals faster. What We Learned We learned that small design tweaks can make a big difference in usability. Also, combining computer vision with live recipe suggestions needs careful attention to speed and accuracy. We realized that showing alternatives and substitutions makes users trust the system more. What’s Next Next, we want to add more ingredients, introduce voice guidance, and make a mobile app. We also plan to send smart reminders for expiring ingredients or missing items. Ultimately, we hope PantryPilot becomes a friendly AI companion in every kitchen.