Inspiration
Like many other people probably, I have black and white photos of my great-grandparents and great-great-grandparents that I wanted to see in color. With my knowledge of machine learning, I sought to create this project.
What it does
Converts any black and white image into a color image
How we built it
Python and PyTorch for the machine learning models (both CNNs) / backend, and Streamlit for the frontend
Challenges we ran into
Streamlit sometimes has issues running large machine learning models, but this was easily fixed with PyTorch. I also had issues trying to deploy where Heroku / Streamlit was installing more packages than what was in my requirements.txt (essentially downloading all packages in my virtual environment), which was more megabytes than either service allowed.
Accomplishments that we're proud of
This was my first time creating a model with PyTorch instead of using Tensorflow/Keras or scikit-learn!
What we learned
How to create a model with PyTorch
What's next for Image Colorizer
Deploy to Heroku/Streamlit, and then add user functionality (save user's images, communicate with other users)
Log in or sign up for Devpost to join the conversation.