Synthetic data generation using viewpoint augmentation for autonomous systems and applications

This invention describes a method for a computer program to generate new views of a scene from different angles. It takes a series of existing images, like frames from a video, and uses one image to predict what the scene would look like from a new perspective. The program then compares its prediction with another actual image from the sequence and uses depth information to refine its ability to create accurate new viewpoints.