Inspiration
Inspired by modern reimaginations of contemporary art. With the existence of AI based art generators such as DALL-E2, we wanted to explore this space of computer generated art throgh the use of ML.
What it does
Our application takes two images a base image and style image in which the style will be applied to the base image.
How we built it
Leverage machine learning in the form of a pretrained convoluted neural net to apply chosen stylizing onto base image across several iterations. The project is primary built using python and more specifically pytorch for machine learning portion. For the front end we use TKinter to construct a basic GUI to support image uploading and displaying of stylized base image.
Challenges we ran into
Initially planned to use reactjs for front end however we faced difficulties in connecting front end interface to backend machine learning model.
Accomplishments that we're proud of
Having a submittable project by the end of the hack, gain insights into machine learning application for stylizing images.
What we learned
Exposure to new technology and framework such as TKinter, and Pytroch
What's next for Down2Style
A react based front end and perhaps even hosting this image style transfer as a webservice via utilizing the cloud.
Log in or sign up for Devpost to join the conversation.