首页>
外国专利>
Deep Stereo: Learning to predict new views from real world images
Deep Stereo: Learning to predict new views from real world images
展开▼
机译:深度立体声:学习根据现实世界的图像预测新的视角
展开▼
页面导航
摘要
著录项
相似文献
摘要
Deep learning systems and methods that use deep networks to predict new views from existing images create and refine models and representations from large data. This deep learning system and method utilizes a deep architecture that performs new view synthesis directly from pixels trained from numerous sets of images. A system using this type of deep network can produce pixels of an invisible view based on the pixels of neighboring views, providing itself to the application when generating graphics.
展开▼