Frame-Fit

• 

• 

• 

  Machine Learning

Inspiration

What ignited this idea was style. In both business and personal life, our appearance is one of the most important things when talking about first impressions.

What it does

Our app scans your facial structure to find the perfect eyewear for you!

How we built it

We used modules like OpenCV, and Google mediapipe to detect the facial landmarks. we then find certain distances on the face using the Euclidian distance formula and a loop. These distances are then recorded in a csv file, where our machine learning model will take this data and detect face shape. it does this through multivariate regression and compares line of best fit to images of celebrities. https://www.kaggle.com/datasets/niten19/face-shape-dataset

Challenges we ran into

We encountered some frustrating challenges. but in the end, the team persevered and overcame them. One challenge included finding the correct facial measurements for the CSV data. another challenge we ran into was connecting the web app to the algorithm, and keeping it speedy.

Accomplishments that we're proud of

we are proud of the amazing effort this team has gone through, creating an app for the first time, with little time, and a huge learning curve.

What we learned

OpenCV mediapipe socket flask Machine Learning Computer Vision Euclidian and Manhattan Distance Linear Regression Multivariate Regression collaboration Git/GitHub Miro

What's next for Frame-Fit

Next, is fine tuning the machine learning algorithm, making it more accurate, and helping people accomplish their styling needs. Next would be to fix up the front end to make it user friendly, and to make the algorithm faster

Built With

• flask
• javascript
• mediapipe
• numpy
• opencv
• pandas
• python
• sklearn
• socket