公开(公告)号：US11715251B2

公开(公告)日：2023-08-01

申请号：US17507620

申请日：2021-10-21

Applicant: NVIDIA Corporation

Inventor： Jonathan Tremblay ,  Aayush Prakash ,  Mark A. Brophy ,  Varun Jampani ,  Cem Anil ,  Stanley Thomas Birchfield ,  Thang Hong To ,  David Jesus Acuna Marrero

IPC: G06T15/00 ,  G06T15/04 ,  G06T15/50 ,  G06T15/20 ,  G06F18/214 ,  G06F18/211 ,  G06V10/774 ,  G06V10/82 ,  G06N3/04 ,  G06N3/084

CPC classification number: G06T15/00 ,  G06F18/211 ,  G06F18/2148 ,  G06T15/04 ,  G06T15/20 ,  G06T15/50 ,  G06V10/7747 ,  G06V10/82 ,  G06N3/04 ,  G06N3/084 ,  G06T2210/12 ,  G06V2201/07

Abstract: Training deep neural networks requires a large amount of labeled training data. Conventionally, labeled training data is generated by gathering real images that are manually labelled which is very time-consuming. Instead of manually labelling a training dataset, domain randomization technique is used generate training data that is automatically labeled. The generated training data may be used to train neural networks for object detection and segmentation (labelling) tasks. In an embodiment, the generated training data includes synthetic input images generated by rendering three-dimensional (3D) objects of interest in a 3D scene. In an embodiment, the generated training data includes synthetic input images generated by rendering 3D objects of interest on a 2D background image. The 3D objects of interest are objects that a neural network is trained to detect and/or label.

公开(公告)号：US20210097346A1

公开(公告)日：2021-04-01

申请号：US17119971

申请日：2020-12-11

Applicant: NVIDIA Corporation

Inventor： Jonathan Tremblay ,  Aayush Prakash ,  Mark A. Brophy ,  Varun Jampani ,  Cem Anil ,  Stanley Thomas Birchfield ,  Thang Hong To ,  David Jesus Acuna Marrero

IPC: G06K9/62 ,  G06T15/04 ,  G06T15/50 ,  G06T15/20

Abstract: Training deep neural networks requires a large amount of labeled training data. Conventionally, labeled training data is generated by gathering real images that are manually labelled which is very time-consuming. Instead of manually labelling a training dataset, domain randomization technique is used generate training data that is automatically labeled. The generated training data may be used to train neural networks for object detection and segmentation (labelling) tasks. In an embodiment, the generated training data includes synthetic input images generated by rendering three-dimensional (3D) objects of interest in a 3D scene. In an embodiment, the generated training data includes synthetic input images generated by rendering 3D objects of interest on a 2D background image. The 3D objects of interest are objects that a neural network is trained to detect and/or label.

公开(公告)号：US11182649B2

公开(公告)日：2021-11-23

申请号：US17119971

申请日：2020-12-11

Applicant: NVIDIA Corporation

Inventor： Jonathan Tremblay ,  Aayush Prakash ,  Mark A. Brophy ,  Varun Jampani ,  Cem Anil ,  Stanley Thomas Birchfield ,  Thang Hong To ,  David Jesus Acuna Marrero

IPC: G06K9/62 ,  G06T15/04 ,  G06T15/50 ,  G06T15/20 ,  G06N3/04 ,  G06N3/08

Abstract: Training deep neural networks requires a large amount of labeled training data. Conventionally, labeled training data is generated by gathering real images that are manually labelled which is very time-consuming. Instead of manually labelling a training dataset, domain randomization technique is used generate training data that is automatically labeled. The generated training data may be used to train neural networks for object detection and segmentation (labelling) tasks. In an embodiment, the generated training data includes synthetic input images generated by rendering three-dimensional (3D) objects of interest in a 3D scene. In an embodiment, the generated training data includes synthetic input images generated by rendering 3D objects of interest on a 2D background image. The 3D objects of interest are objects that a neural network is trained to detect and/or label.

公开(公告)号：US20190251397A1

公开(公告)日：2019-08-15

申请号：US16256820

申请日：2019-01-24

Applicant: NVIDIA Corporation

Inventor： Jonathan Tremblay ,  Aayush Prakash ,  Mark A. Brophy ,  Varun Jampani ,  Cem Anil ,  Stanley Thomas Birchfield ,  Thang Hong To ,  David Jesus Acuna Marrero

IPC: G06K9/62 ,  G06T15/04 ,  G06T15/50 ,  G06T15/20

CPC classification number: G06K9/6257 ,  G06K9/6228 ,  G06K2209/21 ,  G06N3/04 ,  G06N3/084 ,  G06T15/04 ,  G06T15/20 ,  G06T15/50 ,  G06T2210/12

Abstract: Training deep neural networks requires a large amount of labeled training data. Conventionally, labeled training data is generated by gathering real images that are manually labelled which is very time-consuming. Instead of manually labelling a training dataset, domain randomization technique is used generate training data that is automatically labeled. The generated training data may be used to train neural networks for object detection and segmentation (labelling) tasks. In an embodiment, the generated training data includes synthetic input images generated by rendering three-dimensional (3D) objects of interest in a 3D scene. In an embodiment, the generated training data includes synthetic input images generated by rendering 3D objects of interest on a 2D background image. The 3D objects of interest are objects that a neural network is trained to detect and/or label.

公开(公告)号：US20190114546A1

公开(公告)日：2019-04-18

申请号：US16153430

申请日：2018-10-05

Applicant: NVIDIA Corporation

Inventor： Cem Anil ,  Gavriel Robert State

IPC: G06N3/08 ,  G06N3/04

Abstract: A method, computer readable medium, and system are disclosed for identifying predicted labels for a plurality of time-associated data points. The time-associated data points may include data points ordered according to time, where each data point is associated with an individual time value. A convolutional neural network (CNN) is also trained, utilizing the predicted labels for the plurality of time-associated data points as well as ground truth labels for the plurality of time-associated data points, in order to create a trained CNN. The ground truth labels may include labels for the time-associated data points that have been verified as correct. Error correction is then performed, utilizing the trained CNN. For example, the trained CNN may create and return error-corrected labels for another plurality of time-associated data points, based on the predicted labels for the other plurality of time-associated data points, which may increase an accuracy of the labels.