Colour My World
Using a convolutional neural network and a ton of image processing to intelligently colour an artist's sketch
Toronto
Technologies Used
Javascript (Angular / JQuery / Node)About
Using the caffe convolutional neural network framework, our team has created a drawing web app using the Wacom WILL (Universal Ink) API for artists to bring their creations to life with the press of a few buttons in amazing times. Wacom offers a wide range of artist tools that cater towards the most advanced artists and designers. Our tool allows these skilled artists to instantly colourize their sketches using object and feature extraction to intelligently match colours to the sampling pixels on the sketch. This not only adds productivity to the creative process using advanced algorithms in Berkley's deep learning image framework, but also adds a new dimension of creativity to someone's art. It's not everyday that you can say that a computer helped you re-imagine a piece of art. On top of the accurate colourization model, our team added hue, saturation, vibrance, and sampling rate features to give the artist even more freedom in which they colourize and gain inspiration. The best thing about this project is that it is completely online and can possibly be used in the Wacom's Inkspace cloud system. Soon everyone may have access to deep learning at their fingertips. As an extra add on, using feature extraction, we created smart hashtags that automatically post the photo and it's hashtags to the world on twitter at the press of one button. This is a novel way of creating art and could have potential to streamline the entire design process. This is Colour My World
Comments
Christopher Lewis - 2550d
Himel Mondal - 2550d
Himel Mondal - 2550d
David Pellerin - 2550d
Himel Mondal - 2550d
Himel Mondal - 2550d
Varun Venkataramanan - 2550d
Himel Mondal - 2549d
Himel Mondal - 2549d
Himel Mondal - 2549d
Himel Mondal - 2547d