公开(公告)号：US20210166052A1

公开(公告)日：2021-06-03

申请号：US17109421

申请日：2020-12-02

Applicant: NVIDIA Corporation

Inventor： Minwoo Park ,  Yilin Yang ,  Xiaolin Lin ,  Abhishek Bajpayee ,  Hae-Jong Seo ,  Eric Jonathan Yuan ,  Xudong Chen

IPC: G06K9/46 ,  G05D1/00 ,  G06K9/62 ,  G06K9/00 ,  G06N3/08 ,  G06N3/04

Abstract: In various examples, one or more deep neural networks (DNNs) are executed to regress on control points of a curve, and the control points may be used to perform a curve fitting operation—e.g., Bezier curve fitting—to identify landmark locations and geometries in an environment. The outputs of the DNN(s) may thus indicate the two-dimensional (2D) image-space and/or three-dimensional (3D) world-space control point locations, and post-processing techniques—such as clustering and temporal smoothing—may be executed to determine landmark locations and poses with precision and in real-time. As a result, reconstructed curves corresponding to the landmarks—e.g., lane line, road boundary line, crosswalk, pole, text, etc.—may be used by a vehicle to perform one or more operations for navigating an environment.

公开(公告)号：US20230214654A1

公开(公告)日：2023-07-06

申请号：US18174856

申请日：2023-02-27

Applicant: c/o NVIDIA Corporation

Inventor： Minwoo Park ,  Yilin Yang ,  Xiaolin Lin ,  Abhishek Bajpayee ,  Hae-Jong Seo ,  Eric Jonathan Yuan ,  Xudong Chen

IPC: G06V20/58 ,  G06V20/56 ,  B60W60/00

CPC classification number: G06V20/58 ,  G06V20/588 ,  B60W60/001 ,  B60W2420/42

Abstract: In various examples, one or more deep neural networks (DNNs) are executed to regress on control points of a curve, and the control points may be used to perform a curve fitting operation—e.g., Bezier curve fitting—to identify landmark locations and geometries in an environment. The outputs of the DNN(s) may thus indicate the two-dimensional (2D) image-space and/or three-dimensional (3D) world-space control point locations, and post-processing techniques—such as clustering and temporal smoothing—may be executed to determine landmark locations and poses with precision and in real-time. As a result, reconstructed curves corresponding to the landmarks—e.g., lane line, road boundary line, crosswalk, pole, text, etc.—may be used by a vehicle to perform one or more operations for navigating an environment.

公开(公告)号：US11651215B2

公开(公告)日：2023-05-16

申请号：US17109421

申请日：2020-12-02

Applicant: NVIDIA Corporation

Inventor： Minwoo Park ,  Yilin Yang ,  Xiaolin Lin ,  Abhishek Bajpayee ,  Hae-Jong Seo ,  Eric Jonathan Yuan ,  Xudong Chen

IPC: G06N3/08 ,  G06V20/58 ,  G06V20/56 ,  G06F18/23 ,  G06F18/214 ,  G06V10/762 ,  G06V10/764 ,  G06V10/82 ,  G06V10/44 ,  G06V10/26 ,  G06V10/46 ,  G05D1/00 ,  G06N3/045 ,  G06V10/75 ,  G06V10/774 ,  G06V10/94

CPC classification number: G06N3/08 ,  G05D1/0088 ,  G06F18/214 ,  G06F18/23 ,  G06N3/045 ,  G06V10/26 ,  G06V10/454 ,  G06V10/46 ,  G06V10/757 ,  G06V10/763 ,  G06V10/764 ,  G06V10/774 ,  G06V10/82 ,  G06V10/955 ,  G06V20/582 ,  G06V20/588 ,  G05D2201/0213 ,  G06V10/471

Abstract: In various examples, one or more deep neural networks (DNNs) are executed to regress on control points of a curve, and the control points may be used to perform a curve fitting operation—e.g., Bezier curve fitting—to identify landmark locations and geometries in an environment. The outputs of the DNN(s) may thus indicate the two-dimensional (2D) image-space and/or three-dimensional (3D) world-space control point locations, and post-processing techniques—such as clustering and temporal smoothing—may be executed to determine landmark locations and poses with precision and in real-time. As a result, reconstructed curves corresponding to the landmarks—e.g., lane line, road boundary line, crosswalk, pole, text, etc.—may be used by a vehicle to perform one or more operations for navigating an environment.

公开(公告)号：US20250022217A1

公开(公告)日：2025-01-16

申请号：US18351917

申请日：2023-07-13

Applicant: NVIDIA Corporation

Inventor： Abhishek Bajpayee ,  Sai Krishnan Chandrasekar ,  Xudong Chen ,  Hae Jong Seo ,  Siddharth Kothiyal

IPC: G06T17/00

Abstract: Systems and methods are disclosed that relate to object detection and to generating detected object representations. Sensor data corresponding to a scene may be obtained that may represent one or more objects. A tensor may be generated based at least on the sensor data, where the tensor may represent the one or more objects and may include respective predicted 3D characteristics of the one or more objects. The tensor may be represented in 2D space and may be decoded to generate 3D representations of objects using, for example, one or more curve fitting algorithms.