LOGO⅃: Logo generator that uses GAN model to create thousands of new logos within minutes.

Data Science & Machine Learning Immersive

Cohort 10

Project description

LOGO⅃ is an image generator that produces new logos in a snap. The logo generator is built upon Deep Convolutional Generative Adversarial Network (DCGAN). DCGAN is composed of two competing models, discriminator (D) and generator (G). The function of D is to distinguish if the input image is real or fake, while G has to generate image which is as real as human-designed logo so as to deceive D. With enough real life logos as training images, the GAN model is learning at a tremendous rate until it is able to generate decent logos consistently. Users can then surf through the sea of generated logos as an idea bank, or even directly choose a logo they like for further use.