Instance Map based Image Synthesis with a Denoising Generative Adversarial Network
This work addresses challenges in generating realistic images from semantic layouts for computer vision applications, but it appears incremental as it builds on existing GAN methods with specific modifications.
The paper tackles the problem of enhancing image synthesis quality and handling overlapped object generation from semantic layouts using a denoising GAN framework, achieving improved results on Cityscapes, Facades, and NYU datasets.
Semantic layouts based Image synthesizing, which has benefited from the success of Generative Adversarial Network (GAN), has drawn much attention in these days. How to enhance the synthesis image equality while keeping the stochasticity of the GAN is still a challenge. We propose a novel denoising framework to handle this problem. The overlapped objects generation is another challenging task when synthesizing images from a semantic layout to a realistic RGB photo. To overcome this deficiency, we include a one-hot semantic label map to force the generator paying more attention on the overlapped objects generation. Furthermore, we improve the loss function of the discriminator by considering perturb loss and cascade layer loss to guide the generation process. We applied our methods on the Cityscapes, Facades and NYU datasets and demonstrate the image generation ability of our model.