公开(公告)号：GB2574757A

公开(公告)日：2019-12-18

申请号：GB201913449

申请日：2017-12-08

Applicant: IBM

Inventor： GUO QIANG HU ,  JUN CHI YAN ,  JUN ZHU ,  NING DUAN ,  JING CHANG HUANG

IPC: B60W30/00

Abstract: One or more sensors associated with a vehicle detect a roadway condition of a first roadway and an evasive maneuver is performed by the vehicle to avoid the detected roadway condition. In response to one or more processors determining that the evasive maneuver was successful, a record of the successful maneuver and the roadway condition are stored in a database. Subsequent to storing the record in the database, one or more computers associated with one or more vehicles is trained to execute the evasive maneuver, in response to determining that the one or more vehicles is exposed to the roadway condition experienced by the vehicle.

公开(公告)号：GB2602553A

公开(公告)日：2022-07-06

申请号：GB202116123

申请日：2021-11-10

Applicant: IBM

Inventor： JIAN XU ,  GUO QIANG HU ,  YUAN YUAN DING ,  FAN LI ,  JINFENG LI ,  JUN ZHU

IPC: B07C5/342

Abstract: A method for removing an anomaly 104A, 104B in a collection of material comprises receiving, by a processor, an image 108 of the collection of material having a plurality of objects 106, identifying the anomaly 104A, 104B from the plurality of objects 106, generating a bounding box 110A, 110B for the anomaly 104A, 104B, generating one or more picking points on the anomaly 104A, 104B, the one or more picking points being configured on at least one balance points of the anomaly 104A, 104B and removing the anomaly 104A, 104B from the collection of material based on the one or more picking points preferably via robotic means. The method may further comprise comparing the area of the bounding box 110A, 110B to a threshold area. If the area of the bounding box 110A, 110B is less than a threshold area and the one or more picking points are generated at the center of the bounding box 110A, 110B. If the area of the bounding box 110A, 110B exceeds a threshold area one or more segmentations of the anomaly are computed and analyzed.

公开(公告)号：GB2600587A

公开(公告)日：2022-05-04

申请号：GB202200865

申请日：2020-06-12

Applicant: IBM

Inventor： FAN LI ,  GUO QIANG HU ,  SHENG NAN ZHU ,  JUN ZHU ,  JING CHANG HUANG ,  PENG JI ,  YUAN YUAN DING

IPC: G06K9/62 ,  G06V10/24

Abstract: A method, a device and a computer program product for image processing are proposed. In the method, whether a first image indicates a defect associated with a target object is determined. In response to determining that the first image indicates the defect, a second image absent from the defect is obtained based on the first image. The defect is identified by comparing the first image with the second image. In this way, the defect associated with the target object in the image can be accurately and efficiently identified or segmented.

公开(公告)号：GB2574757B

公开(公告)日：2021-12-29

申请号：GB201913449

申请日：2017-12-08

Applicant: IBM

Inventor： GUO QIANG HU ,  JUN CHI YAN ,  JUN ZHU ,  NING DUAN ,  JINGCHANG HUANG

IPC: B60W30/00

Abstract: One or more sensors associated with a vehicle detect a roadway condition of a first roadway and an evasive maneuver is performed by the vehicle to avoid the detected roadway condition. In response to one or more processors determining that the evasive maneuver was successful, a record of the successful maneuver and the roadway condition are stored in a database. Subsequent to storing the record in the database, one or more computers associated with one or more vehicles is trained to execute the evasive maneuver, in response to determining that the one or more vehicles is exposed to the roadway condition experienced by the vehicle.

公开(公告)号：GB2580855A

公开(公告)日：2020-07-29

申请号：GB202008897

申请日：2018-08-24

Applicant: IBM

Inventor： WEISHAN DONG ,  TING YUAN ,  KAI YANG ,  CHANGSHENG LI ,  JUN ZHU ,  RENJIE YAO ,  PENG GAO ,  CHUNYANG MA

IPC: G06N3/04

Abstract: Systems and methods training a model are disclosed. In the method, training data is obtained by a deep neural network (DNN) first, the deep neural network comprising at least one hidden layer. Then features of the training data are obtained from a specified hidden layer of the at least one hidden layer, the specified hidden layer being connected respectively to a supervised classification network for classification tasks and an autoencoder based reconstruction network for reconstruction tasks. And at last the DNN, the supervised classification network and the reconstruction network are trained as a whole based on the obtained features, the training being guided by the classification tasks and the reconstruction tasks