SketchyGAN: Towards Diverse and Realistic Sketch to Image Synthesis

Abstract: Synthesizing realistic images from human drawn sketches is a challenging problem in computer graphics and vision. Existing approaches either need exact edge maps, or require a database to retrieve images from. In this work, we propose a novel Generative Adversarial Network (GAN) approach that synthesizes realistic looking images from 50 categories including motorcycles, horses and couches. We demonstrate a data augmentation technique for sketches which is fully automatic, and we show that the augmented data is helpful to our task. We introduce a new building block suit for both the generator and discriminator which improves the information flow and utilizes input images at multiple scales. Compared to state-of-the-art image translation methods, our approach generates more realistic images and achieves significantly higher Inception Scores.