Food Genie

How it Looks on Bootup
After Adding Images to Classify
API Data Acquired
Able to Guess Banana From Given Image

Inspiration

How many times have your promised yourself you would eat healthier and get fit? You broke that promise many times because you couldn't decide on what to eat with what you have, or what you could make with ease?

This is exactly where Food Genie comes in. It provides users with a simple interface to get healthy recipes based on the ingredients they have. All it takes is one click of a button to upload your images and one more to process your data!

What it does

To go into a bit more depth, we trained a machine learning model to classify food items in the image that user uploads. The data obtained is then sent to an API (Spoonacular) to receive recipes based on those food items. After the API data is fetched, our program displays data of recipes you could make with the given items.

How we built it

First, we split our team into frontend and backend with 2 people on each sub team. The frontend is simple and we did not need to use any framework since the whole interface is on one webpage and all the changes that happen are using DOM Manipulation on a image box. It was built in base HTML, CSS, and Javascript. The backend consists of our machine learning model trained to classify food items, and our script to communicate with the API.

Challenges we ran into

Three members of this team do not go to McMaster so they had to commute which meant we started pretty late on Saturday. Our team had no experience with machine learning prior to this hackathon. The biggest challenge was piecing together all the technology required to make this happen, breaking the entire process into tangible steps, and then actually implementing the code for it....

Accomplishments that we're proud of

We are extremely proud of our sleek and simple UI design that makes use of basic web development principles to offer a responsive design. Also, we managed to train a machine learning model to detect food items with a 95% accuracy.

What we learned

This project allowed us to get our feet wet with machine learning. We learned how to feed datasets to a machine learning algorithm in order to train it. We learned the importance of cleaning the dataset before feeding it to our algorithm. In addition, we sharpened our DOM Manipulation skills by forcing ourselves to use one webpage and making the requested changes on that.

What's next for Food Genie

We wanted to add both a text input option and filters so that users can enter the macros they desire and get recommended recipes and meal plans based on that. We also plan to allow the website to produce meal plans for weight loss and other fitness objectives! The idea we came up with is simply the root of a tree that has many branches!