Pantry Minder

Putting a ‘Best Before’ date on those without one

Comment

Inspiration

We noticed that fresh fruits and vegetables rot within a couple of days and find our families throwing them out very often. After all, the ‘best before’ date for fruits and vegetables aren’t labelled. With the rise of consumerism, the increasing business of our lives, and the limitations of determining the best before date of fruits and vegetables, we easily fall into the trap of mismanaging our pantry and wasting fresh food. Pantry Minders is an elegant solution to keep track of the number of days before fruits and vegetables rot, econonomize user’s spending, and most importantly, minimize food waste at source.

What it does

Utilizing a remote machine learning food recognition API, Pantry Minder allows users to instantly generate a list of the fresh produce in their pantry. Pantry Minder’s abundant database also notifies users of each product’s shelf life, allowing them to prioritize which foods to consume before it goes rotten.

How we built it

Pantry Minder is simple for busy moms and university students to use. Users just need to upload a photo of their newly stocked pantry onto Pantry Minder. Then they are notified of each product’s shelf life via a clearly generated list.

Challenges we ran into

We used visual studio code as the code editor and react native expo to render out the app on both iOS and android. We implemented Clarafai’s food recognition API to our food recognition feature.

Accomplishments that we're proud of

The first challenge that we faced was that one of us was unable to run metro bundler, which meant that the code could not be tested locally. Fortunately, we were able to find an alternative that can allow us to test our codes and see how the application looks on a mobile phone, which is using snack.expo.dev to test our codes online. In addition, our first food recognition implementation – Tensorflow’s machine learning library – was too slow on the mobile phone. Fortunately, Clarafai’s remote API provided us with a time-efficient solution for a better user experience.

What we learned

It is our first time using React Native to build a project together and it is also the first-ever hackathon we have ever attended. As such, we are very proud of what we can showcase.

What's next for Pantry Minder

The accuracy of our food recognition API could be improved with more training data and better training algorithms. For now, the database we used can only display the average number of days before a type of fruit is ripe across a range. We hope to train our own machine learning algorithm so that the application can recognize the ripeness of each fruit and its exact shelf life. Moreover, we would like to push our machine learning image recognition technology to be implemented in supermarkets. Further implemented to supermarkets, The freshness of fruits and veggies would be scanned automatically during checkout, and users who sign in to their account at the checkout can instantly get their list of expiry dates.

Share this project:

Updates