One late night, standing before a nearly bare refrigerator except for some noodles, an idea sparked: what if there was a way to connect with others in a similar predicament and pool resources to create a culinary masterpiece together? From this simple yet revolutionary idea, "potluck" was born, bridging the gap between solitary meals and delightful dining experiences shared with new friends.

potluck is an innovative app designed to revolutionize the way we approach cooking. Users can simply take a picture of the contents of their fridge, and through the power of a fine-tuned YOLOv8 object classification model, the app identifies all available ingredients and populates them into a database.

When users look for someone to cook with, potluck kicks into action, sifting through its user base to find other potential cook-mates in their vicinity. Through a Tinder-style interface, they can accept or reject cooking connections based on available ingredients and possible recipes generated through NLP techniques.

potluck stands tall on the robust shoulders of several cutting-edge technologies:

• YOLOv8: Leveraged for object classification to identify ingredients from pictures swiftly and accurately.
• CockroachDB: Ensures a resilient and consistent database.
• Prisma: Facilitates easy database management.
• Python & FastAPI: Forms the backend microservice that calls the YOLOv8 model and sends the detected ingredients to the database.
• tRPC: Enables end-to-end typesafe APIs, enhancing the performance of potluck.
• Cohere: Powers the NLP feature that aids in generating possible recipes.
• Next.js: Builds the frontend, offering a fast and simple user interface.

Developing a streamlined object classification process posed quite a challenge. The YOLOv8 model required meticulous training and fine-tuning to accurately identify a wide variety of food items from different angles and lighting conditions. Integrating several technologies into a seamless, user-friendly experience was another challenging task, demanding synergized efforts and expertise in various tools and platforms.

• Crafting a user-friendly app that transforms solitary cooking into a social experience.
• Successfully integrating cutting-edge technologies to build a multifaceted yet seamless application.
• Developing an object classification model with high accuracy and reliability.

During our development journey, the team delved deeper into the workings of object detection, familiarizing ourselves with the intricacies of YOLOv8. We honed our skills in backend/microservice development, learning to integrate multiple technologies harmoniously to create a solution that is more than the sum of its parts.

Looking forward, we envision augmenting potluck with features such as:

• A rating system for user experiences and recipes.
• Integration with social media platforms for a richer, integrated social experience.
• Enhanced recipe recommendations using machine learning to personalize suggestions based on user preferences and dietary restrictions. We aim to make potluck not just an app, but a community of food enthusiasts where passion for cooking meets collaboration and innovation!