Inspiration

I'm hosting my own vegan recipe website, which uses an automatic ingredient classifier when a recipe is uploaded. I saw a great synergy for this hackathon!

What it does

Using an open source model from Clarifai, it maps the classified ingredients to the desired Picnic categories

How I built it

Clarifai, Cloudinary (for hosting the images online, needed for Clarifai) and of course Python (with the amazing tool Jupyter notebook

Challenges I ran into

The food model from Clarifai is not tuned for packaged food, so this kind of classification is hard. Also rotten food is hard to classify. For instance it will think that a rotten baguette is a fish...

Accomplishments that I'm proud of

Simple MVP, no need for training a complicated model myself, simply combining existing stuff ;)

What I learned

Don't make stuff to overcomplicated!

What's next for picnic_veganwinners_synergy

Combining this pretrained model with a model that is able to recognize packaged food / rotten food better

Built With

  • jupyter-notebook
Share this project:

Updates