As the name says, it rates memes and it does an awful job while at it.
Please check the Awful meme rater.pdf
document to get more in-depth
details than you would get here.
Creating the dataset requires Tesseract to use the python library.
It has two files and each one does their thing. Remember, KISS (keep it simple, stupid)
generate_dataset.py
:
- Gets memes from reddit.com/r/memes
- Stores the images and a
.attrs
file (detailed in the.pdf
)
kmeans.py
:
- Applies k-means to fit memes to
popular
andunpopular
clusters - You can add your own meme with your own .attrs file and see if you're gonna reach hot