Inspiration
Although photo editing/sharing services like Instagram have a massive number of users, there is little to no control over the available filters. We aimed to provide a two click system that would allow users gain this power .
What it does
Our project allows users to create their own personal “Instagram” filters. Basically, users input a photograph to extract artistic style from (or choose from one of our recommended images) and an image to apply this custom filter to. The idea is based on the findings in “A Neural Algorithm of Artistic Style” [2015]. The authors describe a method to separate image style from image content. Effectively, the algorithm iteratively learns more about the content of the image, and batches in the convolutional neural network correspond to some notion of style (e.g. paint strokes, color palates, textures, etc.)
Built With
- amazon-web-services
- azure
- cuda
- deep-learning
- machine-learning
- neural-networks
- python
Log in or sign up for Devpost to join the conversation.