LE SSERAFIND

Home
File Uploaded
Classification

Inspiration

We're both interested in Kpop groups, and have the shared experience of being overwhelmed trying to figure out who's who at first.

What it does

When users input an image of a member into LE SSERAFIND, the program tries to identify what member it is, and outputs their name.

How we built it

The classification model is a small convolutional neural network built using PyTorch. The website itself is React-based, and mainly consists of a "Choose file" button. When a file is selected, it's sent to the CNN and the final classification is returned and displayed.

Challenges we ran into

We collected all of the data on our own, which meant that we weren't able to collect a significant amount of data. Because of this, our model struggled with overfitting and didn't have a particularly high accuracy on the test set.

Accomplishments that we're proud of

We're both usually the designers instead of the programmers, so we're proud of ourselves for taking on such a big task, achieving an accuracy better than guessing, and dipping our toes into full-stack development.