Srinivas Rao

Title: Innovations of Srinivas Rao in 3D Surface Mapping and Image Generation

Introduction

Srinivas Rao is an accomplished inventor based in San Francisco, CA. He has made significant contributions to the fields of 3D surface mapping and image generation, holding two patents that showcase his innovative approach to technology.

Latest Patents

One of his latest patents is titled "Weak multi-view supervision for surface mapping estimation." This invention involves analyzing one or more two-dimensional images of a three-dimensional object to estimate a three-dimensional mesh representing the object. The process begins by determining a correspondence between the images and a UV representation of a three-dimensional template mesh through training a neural network. The three-dimensional template mesh is then deformed to represent the object accurately. This method includes a reprojection loss cycle, where points from the images are mapped onto the UV representation, then onto the three-dimensional template mesh, and finally back onto the two-dimensional images.

Another notable patent is "Deferred neural rendering for view extrapolation." This invention generates novel images using an image generator implemented on a processor. The image generator receives neural features from a neural texture atlas and position guides that identify position information for input image pixels. The generated novel images are evaluated using an image discriminator to determine optimization values by comparing them with corresponding input images. Each novel image is produced from a camera pose relative to an object identical to that of the corresponding input images. The image generator and neural features are updated