Recipe Vault | Devpost

Landing page for our website
Sample image taken by our PiCamera module.
Rough view of our technologies/hardware
2 hours into Knight Hack
36 hours into Knight Hack

Inspiration

Do you find yourself at home constantly asking, "What's there to eat?" Cooking can be hard. Especially when there are so many recipes out there and you don't have all of the ingredients in your kitchen. As broke engineering students, we understand the struggle. We were inspired to develop software that eases the stress of figuring out what to make to eat, while eliminating the need to buy more groceries. Ultimately saving our wallets from eating out and our health from eating instant ramen.

What it does

Recipe Vault uses a food scanner to scan the items in your fridge or pantry. Using an API it compiles a list of the ingredients you have and lists them on the screen. From there, the user can click a button to access recipes that use the ingredients on their list.

How we built it

Using a raspberry pi and an IoT camera module, we were able to create a food scanner that takes images of the food and runs it through Google Cloud Vision. Google Cloud Vision utilizes image recognition software to identify objects in a picture. From there, we used an API to compile the list of ingredients shown in the picture and filter out any irrelevant labels. Using the list of ingredients, we used the Spoonacular API to retrieve recipes using the ingredients from the list.

Challenges we ran into

The most challenging part about this project was trying to accurately scan the images taken by the camera module. We got the camera for $20 on Amazon, and although the quality seems okay, Google Cloud Vision did not like it. The colour and lighting were off, so some of the objects wouldn't be identified correctly. In addition, we also had challenges with the compatibility between our Raspberry Pi and the glibc versions. It was a sensitive dependency that required us to pivot into a different framework.

Accomplishments that we're proud of

Our biggest accomplishment would be figuring out how to use a camera module to analyze data. Most of us have never worked with image recognition technology before and it was fun and exciting to see it all come together in 36 hours. As a team, we were able to help one another and teach different aspects of the project to each other. Even though not all of us were extremely knowledgeable in coding, we used our strength and assets in order to come together and create this awesome product.

What we learned

Some of our members were still quite new to coding, so they were able to get a grasp on JavaScript and website development. We were able to step out of our comfort zone, tackle an image recognition project, and explore a new area of software development. We learnt how to connect our software ideas to hardware devices, and use images taken on a camera module and input them into Google Cloud Vision in order to use an API to create an ingredient list.

What's next for Recipe Vault

In order to improve our hack, we want to incorporate authentication, so that users can store and access their favourite recipes whenever they want to. We also want to eventually implement our software into a mobile app for better convenience and ease of use.