more about curate

I have always been fascinated by topics outside of Computer Science like Art History, Philosophy, Literature, and Sports.
curate is my attempt to apply my knowledge of Data Science to Art History to bring to forefront the contributions of several lesser-known artists who have had substatntial contributions in the Art world.
curate is aimed to be an art exploration tool more than an art predictor tool; a tool for art enthusiasts all around the world. On most of the occasions, the results will not match the actual artist or the art movement but what curate aims to do is to guess a lesser-known artist who might have painted the input image. In this way, curate tries to bring to forefront the many artists who have had their contributions to world of art not highlighted enough.
Through this feature, curate will do a style-transfer for you and e-mail you back the results in a few days once it is done processing. This feature can be used to overlay the style of the painting currently shown in the resultant artist's column over the input image.
Here are a few examples of past style transfers:

Input Image (River Avon)

Artist's painting (Boating by Manet)

Stylized image (e-mailed output)

Input Image (Tate Britain)

Artist's painting (The Wanderer Above the Sea of Fog by Caspar David Friedrich)

Stylized image (e-mailed output)

curate is a Convolutional Neural Network made from ResNet-50 architecture. It was then trained on paintings from previous centuries to learn and pick up signs about what makes each movement and artist special based on the painterly techniques.
Reach out to me at and let me know your thoughts about the project.
This project would not have been possible without the valuable feedback from everyone at the Library Innovation Lab during summer of 2018.
Also, icons are made by Freepik from is licensed by CC 3.0