公开(公告)号：US12018969B2

公开(公告)日：2024-06-25

申请号：US17417317

申请日：2019-11-20

Applicant: NEC Communication Systems, Ltd.

Inventor： Hiroki Yokota ,  Masanori Takaoka ,  Noriyuki Aoki ,  Kenji Kono

IPC: G01F22/00 ,  G01B11/22 ,  G06T7/70

CPC classification number: G01F22/00 ,  G01B11/22 ,  G06T7/70

Abstract: A volume measurement apparatus comprises: a point group conversion part converting depth information from a depth sensor into point group data; a raw material heap detection part detecting a point group related to a raw material heap from the point group data, using repose angle information; and a raw material heap volume calculation part that calculates a volume of the raw material heap non-occlusion part based on the point group related to the raw material heap; estimates a volume of the raw material heap occlusion part in the raw material heap, using at least the point group related to the raw material heap; and calculates a volume of the raw material heap, which is a sum of the calculated volume of the raw material heap non-occlusion part and the estimated volume of the raw material heap occlusion part.

公开(公告)号：US11361547B2

公开(公告)日：2022-06-14

申请号：US16770170

申请日：2018-12-05

Applicant: NEC COMMUNICATION SYSTEMS, LTD.

Inventor： Noriyuki Aoki ,  Masanori Takaoka ,  Tetsuya Ito ,  Hiroki Yokota

IPC: G06K9/00 ,  G06V20/40 ,  G06T7/70 ,  G06N20/00 ,  G06K9/62 ,  G06T7/00 ,  G06V10/98

Abstract: An object detection apparatus comprising: an input part configured to input video data, an extraction part configured to extract image data from the video data, and a determination part configured to determine whether or not the image data includes an object, wherein the determination part is configured to: set a plurality of regions in a learning image data such that the object appearing in the learning image data extracted from a learning video data captured in advance extends over at least the plurality of regions, assign a learning teach label to each of the regions set, and determine whether or not the image data includes the object by using a prediction model generated by utilizing the learning teach label assigned to each of the regions set.