公开(公告)号：US20230326194A1

公开(公告)日：2023-10-12

申请号：US17714865

申请日：2022-04-06

Applicant: GM Cruise Holdings LLC

Inventor： Craig Quiter ,  Siddhartho Bhattacharya ,  Mayank Ketkar ,  Raluca Musaloiu-Elefteri ,  Wanlin Yang ,  Sandeep Gangundi

IPC: G06V10/82 ,  G06V10/774 ,  G06N3/08

CPC classification number: G06V10/82 ,  G06N3/08 ,  G06V10/7747

Abstract: The disclosed technology provides methods for training of a convolutional neural network (CNN) to identify or predict its own errors, and then those errors are used as inputs with feature visualization to generate images of scenes associated with those errors. This allows adjustment of a set of labeled training images, and then the adjusted set of labeled training images are used to retrain or further train the CNN. Systems and machine-readable media are also provided.

公开(公告)号：US20230215134A1

公开(公告)日：2023-07-06

申请号：US17568455

申请日：2022-01-04

Applicant: GM CRUISE HOLDINGS LLC

Inventor： Siddhartho Bhattacharya ,  Craig Quiter ,  Mayank Ketkar ,  Wanlin Yang ,  Sandeep Gangundi ,  Raluca Musaloiu-Elefteri

IPC: G06V10/74 ,  G01S17/86 ,  G06F7/02

CPC classification number: G06V10/761 ,  G01S17/86 ,  G06F7/023

Abstract: The disclosed technology provides solutions for finding samples from image data that are similar to failure cases, by constructing N-dimensional vectors of the failure cases. The vectors of failure cases are compared to other image data, with the objective of identifying groups of images that can be labeled. The labeled images are then used to retrain a model. Systems and machine-readable media are also provided.

公开(公告)号：US12230021B2

公开(公告)日：2025-02-18

申请号：US17714865

申请日：2022-04-06

Applicant: GM Cruise Holdings LLC

Inventor： Craig Quiter ,  Siddhartho Bhattacharya ,  Mayank Ketkar ,  Raluca Musaloiu-Elefteri ,  Wanlin Yang ,  Sandeep Gangundi

IPC: G06V10/82 ,  G06N3/08 ,  G06V10/774

Abstract: The disclosed technology provides methods for training of a convolutional neural network (CNN) to identify or predict its own errors, and then those errors are used as inputs with feature visualization to generate images of scenes associated with those errors. This allows adjustment of a set of labeled training images, and then the adjusted set of labeled training images are used to retrain or further train the CNN. Systems and machine-readable media are also provided.