Fresh Inspiration

Custimize your hello fresh subscription by adding meals, from your favourite social media platform to your meal plan. Instead of searching for perfect recepies you just scroll through your feed.

Comment

Inspiration

How often do you scroll through instagram, see food that looks absolutely amazing, but then don't know how to start with cooking this or where to get all the ingredients ? Our solution is simple! Just share the instagram post to our App and we will find a recipe that is close, and you can integrate it in your existing meal prep subscription. No going out to shop for exotic ingredients involved!

What it does

Simply store dishes with two clicks in your Hello Fresh meal planning app, while browsing through your favourite social media platform (Instagram). It will automatically find a fitting recipe, propose the stored food for meal plans and inspire meal planning recommendations for the future.

How we built it

The Frontend for Fresh Inspiration is built using progressive web apps and the share target web api. This enables the app to behave like any native app and easily integrate with native sharing. The APP is hosted on S3 with a CloudFront distribution for speedy access. The backend is built in python using Flask and uses a larger database of recipes (based on Food.com) to provide better and more personalized recommendations.

Challenges we ran into

One of the core challenges when developing Progressive Web Apps lies in Caching, as browsers - and devices in general- really like to cache everything. We also struggled to find a decent recipe database - while acceptable the one we selected suffers from strange recipe names and was without images, as such we used Stable Diffusion for generating food images, leading to some interesting results.

Accomplishments that we're proud of

We support proper native sharing to our app! The API integration with Instagram was not super easy but we made it work. We also quite like the Stable-Diffusion recipe images for a subset of the dataset that we can show in our prototype. Have a look yourself at some of the weird stuff it came up with! We managed to build a reasonable recommender, even though we had to scale down our dataset due to image processing constraints.

What we learned

Caching is a pain and a saint at once. New web standards can be a bit tricky.

What's next for Fresh Inspiration

If you want to try it out, use the link below on a mobile device (Sharing only works on Android with Chrome or maybe potentially Samsung Internet and Windows with Edge and Chrome)

Built With

Share this project:

Updates