Colorization of Logo Sketch Based on Conditional Generative Adversarial Networks

Logo design is a complex process for designers and color plays very important role in logo design. The automatic colorization of sketch great value full challenges. In this paper, we propose new method based on Conditional Generative Adversarial Networks, which can output multiple colorful logos only by providing one sketch. We improve the traditional U-Net structure, adding channel attention spatial skip-connection. addition, generator consists parallel attention-based blocks, images. During model optimization process, style loss function proposed to diversity logos. evaluate our self-built edges2logos dataset public edges2shoes dataset. Experimental results show that generate more realistic images simple sketches. Compared classic networks, generated network are also superior visual effects.