Sandwichinator

• 

  The Sandwichinator WebApp

Inspiration

How do you define a sandwich? According to your definition is hotdog a sandwich? Well, according to Merriam Webster it is and so are sliders and many other food items that you might or might not consider sandwiches yourself (check out the link for more details- https://www.merriam-webster.com/words-at-play/to-chew-on-10-kinds-of-sandwiches/sloppy-joe)!

Expanding on this concept, this project aims to explore the classification of sandwiches and the boundaries to the definition.

What it does

The Webapp determines whether an image classifies as a sandwich.

How we built it

We downloaded 10,000 images from Google Images and 20,000 images from the Food101 dataset to train a binary classification algorithm on sandwiches using SqueezeNet Network in DeepLearning4J.

Challenges we ran into

We had to switch from using Tensorflow and Python to using DeepLearning4J and Java because we wanted to do everything in memory of the server of the Webapp but the backend of the Webapp is in Java.

What's next for Sandwichinator

Built With

• apachecommonsio
• apachefileupload
• deeplearning4j
• googlecloudapi
• java
• primefaces
• python
• tensorflow
• tomcatserver