公开(公告)号：US20240259694A1

公开(公告)日：2024-08-01

申请号：US18427641

申请日：2024-01-30

Applicant: DENSO CORPORATION

Inventor： Kenji NAKAMURA ,  Kazuhiko KOBAYASHI

IPC: H04N23/73 ,  G06V10/60 ,  G06V20/58 ,  H04N23/61

CPC classification number: H04N23/73 ,  G06V10/60 ,  G06V20/582 ,  G06V20/584 ,  H04N23/61 ,  G06V2201/07

Abstract: An imaging control apparatus is configured to detect a first object and a second object appearing on an image captured by a camera. The first object is higher in luminance than the second object. When the second object is not detected on the image, the imaging control apparatus works to adjusts an exposure time of the camera to a first exposure time which is used when the first object is detected. Alternatively. When the second object is detected, the imaging control apparatus adjusts the exposure time of the camera to a second exposure time which is longer than the first exposure time. This eliminates a failure in identifying colors on the image and also minimizes a risk of appearance of flicker on the image.

公开(公告)号：US20240259629A1

公开(公告)日：2024-08-01

申请号：US18103043

申请日：2023-01-30

Applicant: Adeia Guides Inc.

Inventor： Johan Kölhi ,  Anthony Friede ,  Reda Harb

IPC: H04N21/433 ,  G06V10/764 ,  H04N21/44 ,  H04N21/472

CPC classification number: H04N21/4334 ,  G06V10/764 ,  H04N21/44004 ,  H04N21/47217 ,  G06V2201/07

Abstract: A media stream is accessed, and a classification model is selected for identifying objects in the media stream. Using the classification model, it is determined whether a content item of interest to the user is currently being received in the media stream. If the content item is currently being received in the media stream, at least a portion of the content item is recorded from the media stream. If the content item is not currently being received, then recording is paused or delayed until it is determined that the content item is being received.

公开(公告)号：US20240259525A1

公开(公告)日：2024-08-01

申请号：US18560753

申请日：2021-06-08

Applicant: Hewlett-Packard Development Company, L.P.

Inventor： PETER SIYUAN ZHANG ,  PEI HSUAN LI ,  HSIN CHIN LIN

IPC: H04N7/15 ,  G06T7/70 ,  G06V10/74

CPC classification number: H04N7/15 ,  G06T7/70 ,  G06V10/761 ,  G06T2207/20081 ,  G06T2207/30201 ,  G06V2201/07

Abstract: An example system includes a first image sensor, a second image sensor, and an electronic device. The electronic device includes a controller that is to compare positions of a human face and a non-human object in a first set of images captured by the first image sensor. In addition, the controller is to select a presentation arrangement of the first set of images and a second set of images captured by the second image sensor based on the comparison. Further, the controller is to output the first set of images and the second set of images to a video conferencing application in the selected presentation arrangement.

公开(公告)号：US20240257562A1

公开(公告)日：2024-08-01

申请号：US18626162

申请日：2024-04-03

Applicant: Tencent Technology (Shenzhen) Company Limited

Inventor： Yafei YUAN ,  Wen GE ,  Lulu JIAO ,  Jiayu HUANG ,  Runzeng GUO ,  Ruixin ZHANG ,  Yingyi ZHANG ,  Hang ZHOU ,  Jun WANG

IPC: G06V40/60 ,  G06T7/11 ,  G06T7/62 ,  G06T7/70 ,  G06V10/25 ,  G06V10/74 ,  G06V40/12

CPC classification number: G06V40/67 ,  G06T7/11 ,  G06T7/62 ,  G06T7/70 ,  G06V10/25 ,  G06V10/761 ,  G06V40/1347 ,  G06V2201/07

Abstract: This application discloses a palm image recognition method performed by a computer device. The method includes: performing palm detection on a palm image captured by a camera to generate a palm box for a palm in the palm image (304); determining location information of the palm relative to the camera based on the palm box and the palm image; and displaying a palm identifier corresponding to the palm based on the location information, the palm identifier being used for indicating the palm to move to a preset spatial location corresponding to the camera to obtain an object identifier corresponding to the palm image. A user, based on the palm identifier, moves its palm to the preset spatial location corresponding to the camera to perform palm image recognition.

公开(公告)号：US20240257384A1

公开(公告)日：2024-08-01

申请号：US18424785

申请日：2024-01-27

Applicant: Aptiv Technologies AG

Inventor： Moritz Luszek ,  Simon Roesler

IPC: G06T7/70 ,  G06T7/62 ,  G06V10/764

CPC classification number: G06T7/70 ,  G06T7/62 ,  G06V10/764 ,  G06V2201/07

Abstract: A computer-implemented method for training a birds-eye-view (BEV) object detection model includes inputting a training sample into the model. The training sample includes a BEV image with multiple pixels, and multiple target confidence values. Each pixel of the pixels is associated with a target confidence value of the target confidence values. The method includes receiving as output from the model multiple predicted confidence values. Each predicted confidence value is associated with a pixel of the pixels. The method includes adjusting a parameter set of the model according to a loss. The loss is based on the predicted confidence values and the target confidence values.

公开(公告)号：US20240242432A1

公开(公告)日：2024-07-18

申请号：US18560159

申请日：2022-05-17

Applicant: Robert Bosch GmbH

Inventor： Norman Haag ,  Jan Sparbert ,  Stefan Spiessberger

IPC: G06T17/00 ,  G01S17/89 ,  G06T7/62 ,  G06T7/70 ,  G06V10/26

CPC classification number: G06T17/00 ,  G01S17/89 ,  G06T7/62 ,  G06T7/70 ,  G06V10/26 ,  G06V2201/07

Abstract: A method and a device for determining concealed objects in a 3D point cloud representing an environment. The method includes: producing a 3D point cloud using an active sensor, wherein each point of the 3D point cloud represents a distance measurement by the active sensor; determining a background region within the 3D point cloud; determining shadows within the background region of the 3D point cloud, shadows being the regions within the background region at which there are no points of the 3D point cloud; and determining, in the environment of the sensor, an object which is concealed by crosstalk in the sensor, by identifying, in the background region of the 3D point cloud, at least one shadow which cannot be attributed to an object in the foreground of the 3D point cloud.

公开(公告)号：US20240242377A1

公开(公告)日：2024-07-18

申请号：US18276648

申请日：2023-01-30

Applicant: CHENGDU UNIVERSITY OF TECHNOLOGY

Inventor： Shi YI ,  Xinghao CHENG ,  Cheng WANG

IPC: G06T7/73 ,  B25J9/16 ,  G06V10/46 ,  G06V10/50 ,  G06V10/77 ,  G06V10/80 ,  G06V10/82

CPC classification number: G06T7/73 ,  B25J9/1697 ,  G06V10/462 ,  G06V10/50 ,  G06V10/7715 ,  G06V10/806 ,  G06V10/82 ,  G06T2207/10048 ,  G06T2207/20064 ,  G06T2207/20084 ,  G06V2201/07

Abstract: The present disclosure belongs to the technical field of artificial intelligence, and provides a pose estimation apparatus and method for a robotic arm to grasp a target based on monocular infrared thermal imaging vision. The present disclosure provides a pose estimation apparatus for automatically grasping a target using a robotic arm of a specialized robot, which includes a monocular infrared thermal imaging camera, a laser ranging module, an edge computing platform, and an embedded processor. The monocular infrared thermal imaging camera can perform imaging based on a temperature difference between a target and an environment in smoke, combustion, and explosion environments in which a specialized robot works. The laser ranging module can assist in measuring a distance of a to-be-grasped target. An algorithm process in the present disclosure is deployed on the edge computing platform. The embedded processor completes motion control of a robotic arm. The apparatus is characterized by a lightweight design and high work efficiency, and is suitable for being mounted on a specialized robot platform.

公开(公告)号：US12039874B2

公开(公告)日：2024-07-16

申请号：US17680304

申请日：2022-02-25

Applicant: SZ DJI TECHNOLOGY CO., LTD.

Inventor： Kaiqiang Deng ,  Jiabin Liang ,  Weiwei Zhang ,  Sheng Huang ,  Yinzhuo Wang

IPC: G06T7/00 ,  B64C39/02 ,  G06T7/70 ,  G06V20/10 ,  G06V20/17 ,  G08G5/00 ,  H04N23/69 ,  B64U101/30

CPC classification number: G08G5/006 ,  B64C39/024 ,  G06T7/70 ,  G06V20/17 ,  G06V20/176 ,  G08G5/003 ,  G08G5/0069 ,  G08G5/0086 ,  H04N23/69 ,  B64U2101/30 ,  B64U2201/10 ,  G06T2207/10028 ,  G06T2207/10032 ,  G06T2207/30184 ,  G06T2207/30252 ,  G06V2201/07

Abstract: A control method and device for an unmanned aerial vehicle, and a non-transitory computer readable storage medium are provided. The control method may include: acquiring shooting control information, the shooting control information including a height of a shooting target object, a mission height, and an overlapping ratio; determining the target flight route and a shooting interval based upon the shooting control information; and controlling an unmanned aerial vehicle to fly according to the target flight route and to photograph the shooting target object according to the shooting interval by using a photographing device during the flight. The mission height may be a distance between a plane where a target flight route is located and a plane where the shooting target object is located, and a starting plane for computing the overlapping ratio may be the plane where the shooting target is located.

公开(公告)号：US12039767B2

公开(公告)日：2024-07-16

申请号：US17742735

申请日：2022-05-12

Applicant: GUANGDONG OPPO MOBILE TELECOMMUNICATIONS CORP., LTD.

Inventor： Yuhu Jia

IPC: G06V10/26 ,  G06T7/215 ,  G06V10/25

CPC classification number: G06V10/267 ,  G06T7/215 ,  G06V10/25 ,  G06V2201/07

Abstract: A subject detection method is disclosed. A moving object is detected on an image to obtain a moving object region, candidate regions are obtained by sliding a sliding box on the image, a first region is determined. The first region is one of candidate regions and contains a part of the moving object region with a largest area. A proportion of the first region is obtained, a size of the first region is adjusted based on the proportion to obtain a second region, a proportion of the second region is obtained. The first region is replaced with the second region, the proportion of the first region is replaced with the proportion of the second region, and it is returned to adjusting the size of the first region until a number of iterative adjustments reaches a threshold, a region obtained by the last adjustment is determined as a target region.

公开(公告)号：US20240236436A9

公开(公告)日：2024-07-11

申请号：US18483482

申请日：2023-10-09

Applicant: SHANGHAI BILIBILI TECHNOLOGY CO., LTD.

Inventor： Jun LIU ,  Yicheng ZHU ,  Jingping ZHAO ,  Yu ZHOU

IPC: H04N21/4788 ,  G06V10/25 ,  H04N21/2187 ,  H04N21/2343 ,  H04N21/2743

CPC classification number: H04N21/4788 ,  G06V10/25 ,  H04N21/2187 ,  H04N21/234363 ,  H04N21/2743 ,  G06V2201/07

Abstract: A bullet-screen comment display method is provided. The method includes: detecting a target object in an original frame image of a video at a first preset frequency in a process of playing the video, and when the target object is detected, outputting a display area of the target object in the original frame image; generating a mask image corresponding to the original frame image based on the display area, obtaining a bullet-screen comment display area corresponding to the original frame image; and if there is an overlapping area between the bullet-screen comment display area and the display area of the target object in the original frame image, combining the mask image and a bullet-screen comment displayed in the bullet-screen comment display area into a video picture with a non-obstructive bullet-screen comment, and displaying the video picture with a non-obstructive bullet-screen comment at a bullet-screen comment layer, where a bullet-screen comment included in the video picture with a non-obstructive bullet-screen comment does not block the target object.