公开(公告)号：US20250095229A1

公开(公告)日：2025-03-20

申请号：US18397081

申请日：2023-12-27

Applicant: NVIDIA Corporation

Inventor： Yue Wang ,  Jiawei Yang ,  Boris Ivanovic ,  Xinshuo Weng ,  Or Litany ,  Danfei Xu ,  Seung Wook Kim ,  Sanja Fidler ,  Marco Pavone ,  Boyi Li ,  Tong Che

IPC: G06T11/00 ,  G06T17/00 ,  G06V10/44 ,  H04N13/279

Abstract: Apparatuses, systems, and techniques to generate an image of an environment. In at least one embodiment, one or more neural networks are used to identify one or more static and dynamic features of an environment to be used to generate a representation of the environment.

公开(公告)号：US20240273802A1

公开(公告)日：2024-08-15

申请号：US18422650

申请日：2024-01-25

Applicant: NVIDIA Corporation

Inventor： Yue Wang ,  Marco Pavone ,  Jiawei Yang

IPC: G06T15/00 ,  G06T15/06

CPC classification number: G06T15/005 ,  G06T15/06

Abstract: In various examples, frequency regularization and/or occlusion regularization techniques may be used to train Neural Radiance Fields (NeRF) to determine neural renderings based at least on sparse inputs in a way that reduces overfitting, underfitting, and/or occlusions. For example, while training a NeRF, a linearly increased frequency mask may be applied to regularize a visible frequency spectrum of training data based on training time steps. In examples, as training of the NeRF progresses, the visible frequency may be increased in a way that reduces the risk of overfitting and/or avoids underfitting. Additionally, the disclosed techniques may also include masking one or more density scores located within a threshold proximity of an origin of a ray to reduce floaters, walls, and other occlusions in the neural rendering output.