FridgeFocus

Inspiration

As university students, we often turn to the internet for cooking inspiration. But more often than not, the recipes we find don’t perfectly match the ingredients we already have at home. This usually means making a quick trip to the convenience store to buy a missing item—only for it to end up forgotten and eventually expire in the fridge. To solve this, we came up with an app that tracks the ingredients we already have and generates recipe ideas using only what's available in our fridge!

What it does

FridgeFocus lets you snap a photo using your mobile device, then automatically detects the ingredients shown in your fridge or image. These ingredients are added to your inventory, with quantities specified by you. When you generate a recipe and mark it as "Cooked," the app will automatically update your inventory by subtracting the ingredients used in that recipe. It also allows personal customization that accept request from the user.

How we built it

For the front-end, we initially designed the app using Figma. We then implemented the design using the Kotlin language to construct an Android-app.

For the AI model, we used the OpenCV and YOLO library to train a dataset that contains different types of ingredients. We then extracted the result and transferred the detected object to our backend, where we will store the inventory that the user have.

We used Amazon EC2's RDBS to construct the database, before using Python and Flask for the backend and constructing the REST APIs. As we aim to also run the application by using Android mobile device, so we opted to render our back end using Render.com. It is now running live.

Challenges we ran into

As each of our members have their own expertise, it was really hard to connect the client and the server even until the very end. We have implemented the UI, AI, and back end successfully, but we struggled with connecting everything together. This might be due to the lack of understanding of each other's tech stack, and the under estimation of being able to connect everything without prior investigation.

Accomplishments that we're proud of

However, we are very proud of our idea and work so far. We have pushed our boundaries by learning tech stacks that we are all not familiar about, and although we didn't achieve the perfect result, we still pushed through to submit the project.

What we learned

It's very important to learn about each other tech stack, or at least have a little understanding of what each other is doing. We encountered multiple miscommunications, and there's a confusion in the responsibility of task. We will surely bring this lesson to our future hackathons and learn a lot from this.

What's next for FridgeFocus

Connecting the back-end and front end! More research. We are also thinking about a feature that allows people to scrape prices from different supermarket to get the best deals, further supporting the problem statement of buying only necessary items with low cost without having to throw away anything.

Built With

• amazon-web-services
• figma
• flask
• kotlin
• opencv
• python
• rdbs
• render
• yolo