基于图像条件的二元合成生成算法

生成对抗网络(GAN)是一种无监督学习方法,该算法巧妙地利用博弈的思想来学习生成式模型,但由于GAN通常从单个潜在源采样,因此常常丢失场景中的多个实体交互信息.为了捕获不同对象之间的复杂交互,包括它们的相对缩放,空间布局,遮挡,提出了一种基于图像条件的生成对抗网络,利用分解—合成的流程,模型可以根据输入对象的纹理和...

Binary Synthesis Generation Algorithm Based on Image Conditions

The Generative Adversarial Network (GAN) can generate images with significant complexity andauthenticity, but it is usually constructed to sample from a single potential source, thus ignoring spatial interactionsbetween multiple entities that may exist in the scene. To capture the complex interactions between different objects,including their relative scaling, spatial layout, occlusion or view conversion, in this paper we propose an imagecondition-based generative adversarial network by using decomposition-synthesis procedures. The model cangenerate realistic composite images from their joint distributions based on their texture and shape of the input object.Through use of Shapenet data set, we experiment on 51 300 image models out of the respective 55 common objects inthe 2D and 3D images. Compared with the traditional SLP and cGAN, the image quality in our algorithm can beimproved by 4%.