This project is an earlier version of Hisham's "meme generator" repository. It is left on my page to display the commit history that I (@rishitkhare) am accredited with. This commit history was lost when migrating the repository to a Hisham's account. For the fully up-to-date source code of the site, please visit https://hishambhatti.github.io/meme-generator . The rest of this README is directly lifted from the now updated project repo as of 02/13/2026.
This project uses deep learning and computer vision to create a meme caption for a user-inputted image.
Website available at https://hishambhatti.github.io/meme-generator
Designed by Hisham Bhatti, Samarjit Kaushik, Parum Misri, and Rishit Khare.
This repo is an extension to our final project in CSE 599: Deep Learning. During that course, my partners (Samarjit Kaushik, Parum Misri) and I trained and deployed a model that would take an input image and spit out a humorous image. After much training and refining, the end model was funny... though maybe not always original.
Afterwards, I wanted to deploy this model to the Internet so that family and friends (and potential future employers 👀) could view our project and play around with our creation. With the help of Rishit Khare, we have finally deployed this app live on the Internet. So feel free to play around with it!
The details of how our model works internally can be found in this paper! It also discusses some challenges with training and evaluating our model. As for the meme generator frontend, note that at the moment, any picture you send will be cropped from the center to be a square.
Here is a gallery of some of my favorite generated captions:
 |
 |
|---|
 |
 |
|---|
 |
 |
|---|
Languages: Jupyter Notebook, Python, Javascript/Typescript, CSS, HTML Libraries: PyTorch Frameworks: Flask, React, Tailwind CSS Developer Tools: Google Cloud (for backend server), Git, Github
- Creators: Hisham Bhatti, Samarjit Kaushik, Parum Misri, and Rishit Khare (for full-stack deployment)
- Thanks to Prof. Ranjay Krishna and the entire CSE 599 G staff!
- Make a text generate animation
- Fix the cropping issue (pad to square)
- Add a drop feature (in addition to select)
- Make sure download has white background
- Maybe have the website query the backend first to establish a connection