Does the US Judge a Book by its Cover?

Inspiration

We were inspired by how the electoral college vote favors certain demographics. We just did not know which demographic and this project hopes to solve this and give insights to people on how much their vote counts.

What it does

It takes an image from a react front end, into a convolutional neural network trained in pytorch which determines metrics such as age, race, and gender. It then passes this into a function which uses our dataset to determine how many electoral votes the person is worth.

How I built it

We built a full-stack application with a React front end and a flask backend with pytorch integration.

Challenges I ran into

Some of the challenges we ran into include training the the CNN and figuring out if our model works on computers that do not have NVIDIA GPUs.

Accomplishments that I'm proud of

We are proud of training our first neural network from scratch in pytorch and integrating it with a front end.

What I learned

We learned about how to train models in Pytorch and how to deploy them using Flask

What's next for Does the US Judge a Book by its Cover?

We would like to address the fact that our model does not work on machines without NVIDIA GPUs and we would like to make our front end more cohesive.

Teammates: Avi Patel, Aref Malek, Arnob

Built With

• cnns
• flask
• numpy
• python
• pytorch
• react