Being CS students, we saw that a lot of our classmates were a little challenged in the dating department. Wanting to give back to our community here at Waterloo, we decided to start a course Waterloo wouldn't teach: the course of life, and love.
What it does
The University of Tinder provides a set of tools for users to improve their profile photo selections and social skills. Users can upload their photos to the website and an AI-powered algorithm evaluates how attractively the person is portrayed, from 0% to 100%. Additionally, users can converse with our virtual chatbot that, through the guise of a social interaction, provides real-time feedback to attempts to make conversation and establish a connection.
How we built it
The website was built with HTML, CSS (using the Tailwind framework) and vanilla JS. The image evaluation component was written in Python using Tensorflow Keras. A simple Flask HTTP server allowed for communication between the frontend and backend.
Challenges we ran into
While building this project, the greatest challenge came from training the convolutional neural network, particularly with finding enough data to sufficiently reduce the variance in the model. We tried using a plethora of data augmentation techniques, as well as using ADAM and dropout, but found that even so, the lack of varied data resulting in high variance could not be perfectly remedied.
Accomplishments that we're proud of
We're highly satisfied that we were able to complete the entire project in the given time frame, especially considering that we were new to Tailwind prior to building this website.
What we learned
Web dev is no joke!
What's next for UTinder