Meme Classifier

Inspiration

I have so many memes just sitting on my phone and I can never find the right one when I want to show someone a specific meme. With this program it will allow you to search for a specific meme template and then make it easier to find the meme to fit your need.

What it does

It takes an input of memes and then outputs a folder full of categorized memes.

How I built it

I made a python script to download a database of images. Then I used Jupyter Notebooks with Tensorflow and Keras to make an algorithm that has the ability to do image recognition to classify which meme template is being used in the image.

Challenges I ran into

Making a large enough database that will allow me to train the model properly.

Accomplishments that I'm proud of

I spent a long time making a script to scrape about 30,000 images from a website into certain organized folders to created a database to use to train my algorithm. I also was able to program an image recognition algorithm that can decipher the difference between two images. That is all I had time for, but this can be expanded to implement more meme templates.

What I learned

I learned a lot from both parts of this project. Before this event, I did not have any experience with scripting in Python. However, after this event I feel like I have really strengthened my skills in scripting. I also have not done a lot of work with machine learning in a while, but with this project I began to remember/learn more about this field and why it is o important to our future.

What's next for Meme Classifier

For this project the next steps is to train it with more templates. Then once it is fully trained, then I will be able to pair this algorithm with a front end app and release it to the public for commercial use.