In todays day and age, data is everything. Social media platforms like Snapchat, Instagram and Facebook see hundreds of images of your face each and every day. This provides the opportunity to pull in lots of useful data about you as a person. In the world of targeted advertisements, they utilize search history to predict items that you might be interested in. However, these advertisements don't take into account your physical appearance to provide you with advertisements for items that you actually need.

What it does

JABBIC harnesses the power of machine learning to recognize specific facial qualities (if you wear lots of lipstick, if you are bald, if you have a beard, if you wear glasses, etc.) and utilizes that data to provide better targeted advertisements that actually pertain to who you are as a person.

How we built it

The application itself was built using React Native. We used a subset of a dataset of over 200,000 images to train a machine learning model using AutoML. Python was used to help work with/sort the dataset.

Challenges we ran into

As a team we were all completely new to machine learning. In addition, React Native was new to several of us.

Accomplishments that we're proud of

We had a group member that had never attended a hackathon before QHacks. We are extremely proud of our ability to work as a team and mesh our different backgrounds and disciplines together to make a cool project. We are really proud of everything that we learned and are excited to have a working application at the end of the hackathon.

What we learned

In addition to learning several different programming languages and technologies, we learned the importance of working as a team and having a good time!

What's next for JABBIC

Improving accuracy of the model by training it with the full database of images. Integrating JABBIC with already existing applications such as Snapchat instead of it being a standalone application. In addition, giving JABBIC a voice so that it can describe peoples facial features in images for the visually impaired. Lastly, improve the overall security of the application.

Share this project: