Spark-a-Dish

Inspiration

Spark-a-Dish came to life from a very relatable problem: the stress of trying to cook something new at home. We’ve all felt that excitement about making a new recipe, only to be bogged down by the tedious process of finding all the right ingredients and comparing prices. We wanted to create a solution that would take the hassle out of cooking and bring back the joy.

What it does

Spark-a-Dish is like your personal kitchen assistant. You just upload a photo of the dish you want to make and our system identifies the ingredients in the photo and instantly finds the best options available at Walmart, complete with prices and details. No more endless searching or price comparisons—you can get everything you need with just a few clicks and dive right into cooking.

How we built it

We built Spark-a-Dish with one goal in mind: to make your life easier in the kitchen. It’s a simple web app that takes a dish photo and along with Gemini 1.5 Flash model and our prompt engineering we were able to exactly identify what ingredients the dish needs. Then, using web scraping it scours Walmart’s website in real-time, pulling up product options with the best seller tags along with all the details you need—prices, names and pictures. We wanted to make sure it’s quick and seamless so you can spend more time cooking and less time searching.

Challenges we ran into

Getting the system to accurately identify ingredients from all kinds of dish photos was a big challenge. We also had to make sure our real-time searches on Walmart’s site were smooth as Walmart's API is not open to use for everyone. And of course, we wanted to keep everything fast and user-friendly, even with all that data being processed in the background.

Accomplishments that we're proud of

We’re really proud of how Spark-a-Dish turned out. It’s amazing to see how accurately it identifies ingredient and pulls in product options from Walmart, the way our website looks and feels smooth and responsive and also how accurately it can identify dishes.

What we learned

We learned how to work with image recognition and Gemini's API in a way that feels natural and useful, and we got hands-on experience with real-time data retrieval from different websites and also developing and deploying our very first website on a server.

What's next for Spark-a-Dish

We’re looking to improve response times for our bot as currently we have to wait a specified amount of time before we can pull a product's information from Walmart's website. We will also work on improving the image analysis of out bot to accurately identify dishes from various cultures and countries.