Inspiration
I was inspired by the overpopulation of memes on the internet.
What it does
It applies image processing and machine learning algorithms on all images in your browser to identify any possible memes. Each image is assigned a "dankness" coefficient, and those passing the meme threshold are withheld from your browser's display.
How I built it
I built a machine learning/computer vision API on AWS using Flask, Scikit-learn, and OpenCV. I then built a chrome extension that hooks up to it via HTTP requests.
Challenges I ran into
Building a large dataset of memes/non-memes. Identifying features that distinguish memes from non-memes.
Accomplishments that I'm proud of
Got the classification algorithm to 63% accuracy Created a pretty stellar Google Images scraper in the process
What I learned
How to use scikit-learn, how to make chrome extensions.
What's next for Stanley Memer
We plan to IPO next month
Log in or sign up for Devpost to join the conversation.