中科微点-全球专利检索

  1. Login
  2. Sign Up

Search Results

13 records (took 0.015 seconds)

    IMAGE MODIFICATION TECHNIQUES
    2.
    发明公开
    IMAGE MODIFICATION TECHNIQUES 审中-公开

    公开(公告)号：US20240112404A1

    公开(公告)日：2024-04-04

    申请号：US18480304

    申请日：2023-10-03

    Applicant: QUALCOMM Incorporated

    Inventor： Meng-Lin WU ,  Chung-Chi TSAI ,  An CHEN

    IPC: G06T17/20 ,  G06T3/20 ,  G06T3/40 ,  G06T3/60 ,  G06T7/11 ,  G06T7/50 ,  G06T19/20

    CPC classification number: G06T17/205 ,  G06T3/20 ,  G06T3/40 ,  G06T3/60 ,  G06T7/11 ,  G06T7/50 ,  G06T19/20 ,  G06T2219/2016

    Abstract: Systems and techniques are described herein for modifying the scale and/or position of objects in images. For instance, a system can obtain a two-dimensional (2D) input image from a camera and a three-dimensional (3D) representation of the 2D input image. The system can further determine a first portion of the 3D representation of the 2D input image corresponding to a target object in the 2D input image. The system can adjust a pose of the first portion of the 3D representation of the 2D input image corresponding to the target object. The system can further generate a 2D output image having a modified version of the target object based on the adjusted pose of the first portion of the 3D representation of the 2D input image corresponding to the target object to be output on a display.

    OPPORTUNISTIC POSITION LOCATION DETERMINATION AND REPORTING FOR ASSET TRACKING AND MONITORING
    3.
    发明申请
    OPPORTUNISTIC POSITION LOCATION DETERMINATION AND REPORTING FOR ASSET TRACKING AND MONITORING 有权

    公开(公告)号：US20250119834A1

    公开(公告)日：2025-04-10

    申请号：US18483302

    申请日：2023-10-09

    Applicant: QUALCOMM Incorporated

    Inventor： An CHEN ,  Alex PARK

    IPC: H04W52/02

    Abstract: A method is disclosed herein. The method includes receiving, from a server, first data that is indicative of a first likelihood of the UE successfully performing at least one of a position fix or a transmission of sensor data based on a first set of characteristics associated with the UE and at least one additional UE. The method includes computing, based on the first data and a second set of characteristics associated with the UE, a second likelihood of the UE successfully performing at least one of the position fix or the transmission of the sensor data. The method includes scheduling, based on the second likelihood, a wake-up time instance or a sleep time instance. The method includes transitioning the UE (1) from a sleep state to an active state at the wake-up time instance or (2) from the active state to the sleep state at the sleep time instance.

    ZONE GRADIENT DIFFUSION (ZGD) FOR ZONE-BASED FEDERATED LEARNING
    4.
    发明公开
    ZONE GRADIENT DIFFUSION (ZGD) FOR ZONE-BASED FEDERATED LEARNING 审中-公开

    公开(公告)号：US20240135192A1

    公开(公告)日：2024-04-25

    申请号：US18461410

    申请日：2023-09-04

    Applicant: QUALCOMM Incorporated

    Inventor： Vijaya Datta MAYYURI ,  An CHEN

    IPC: G06N3/098

    CPC classification number: G06N3/098

    Abstract: A processor-implemented method includes receiving machine learning model updates from clients in a federated learning system. The method also includes determining a fixed local zone associated with each of the clients, the fixed local zone having a first fixed boundary. The method includes updating model weights of a central machine learning model based on local machine learning updates for a local subset of the clients corresponding to the fixed local zone. The method includes updating the model weights of the central machine learning model based on neighbor machine learning updates for a neighbor subset of the clients. The neighbor subset corresponds to a fixed neighbor zone that neighbors the fixed local zone and has a second fixed boundary. The neighbor machine learning updates have a different weight than the local machine learning updates when updating model weights. A value of the different weight corresponds to a similarity parameter.

    HARDWARE-AWARE FEDERATED LEARNING
    5.
    发明公开
    HARDWARE-AWARE FEDERATED LEARNING 审中-公开

    公开(公告)号：US20240086699A1

    公开(公告)日：2024-03-14

    申请号：US17941121

    申请日：2022-09-09

    Applicant: QUALCOMM Incorporated

    Inventor： An CHEN ,  Vijaya Datta MAYYURI

    IPC: G06N3/08 ,  G06N3/04

    CPC classification number: G06N3/08 ,  G06N3/0454

    Abstract: A processor-implemented method for hardware-aware federated learning includes receiving, from a server, information corresponding to a first jointly-trained artificial neural network (ANN). A current hardware capability of a device for on-device training of the first jointly-trained ANN is determined. The device transmits an indication of the current hardware capability to the server. In response to the transmitted indication, the device receives information corresponding to a second jointly-trained ANN) from the server. The second jointly-trained ANN is an adapted version of the first jointly-trained ANN generated based on the indication of the current hardware capability.

    IMAGE MODIFICATION TECHNIQUES
    7.
    发明公开
    IMAGE MODIFICATION TECHNIQUES 审中-公开

    公开(公告)号：US20230143034A1

    公开(公告)日：2023-05-11

    申请号：US17524681

    申请日：2021-11-11

    Applicant: QUALCOMM Incorporated

    Inventor： Meng-Lin WU ,  Chung-Chi TSAI ,  An CHEN

    IPC: G06T17/20 ,  G06T3/40 ,  G06T3/20 ,  G06T3/60 ,  G06T19/20 ,  G06T7/50 ,  G06T7/11

    CPC classification number: G06T17/205 ,  G06T3/40 ,  G06T3/20 ,  G06T3/60 ,  G06T19/20 ,  G06T7/50 ,  G06T7/11 ,  G06T2219/2016

    Abstract: Systems and techniques are described herein for modifying the scale and/or position of objects in images. For instance, a system can obtain a two-dimensional (2D) input image from a camera and a three-dimensional (3D) representation of the 2D input image. The system can further determine a first portion of the 3D representation of the 2D input image corresponding to a target object in the 2D input image. The system can adjust a pose of the first portion of the 3D representation of the 2D input image corresponding to the target object. The system can further generate a 2D output image having a modified version of the target object based on the adjusted pose of the first portion of the 3D representation of the 2D input image corresponding to the target object to be output on a display.

    AUTOMATIC CAMERA GUIDANCE AND SETTINGS ADJUSTMENT
    8.
    发明申请
    AUTOMATIC CAMERA GUIDANCE AND SETTINGS ADJUSTMENT 有权

    公开(公告)号：US20210368094A1

    公开(公告)日：2021-11-25

    申请号：US17071971

    申请日：2020-10-15

    Applicant: QUALCOMM Incorporated

    Inventor： Muhua LI ,  An CHEN

    IPC: H04N5/232 ,  G06N20/00

    Abstract: An image capture and processing device captures an image. Based on the image and/or one or more additional images, the image capture and processing device generates and outputs guidance for optimizing image composition, image capture settings, and/or image processing settings. The guidance can be generated based on determination of a direction that a subject of the image is facing, based on sensor measurements indicating that a horizon may be skewed, another image of the same scene captured using a wide-angle lens, another image of the same subject, another image of a different subject, and/or outputs of a machine learning model trained using a set of images. The image capture and processing device can automatically apply certain aspects of the generated guidance, such as image capture settings and/or image processing settings.

    QUANTIZATION-AWARE FEDERATED TRAINING TO ADDRESS EDGE DEVICES HARDWARE CAPABILITIES
    9.
    发明申请
    QUANTIZATION-AWARE FEDERATED TRAINING TO ADDRESS EDGE DEVICES HARDWARE CAPABILITIES 有权

    公开(公告)号：US20250086426A1

    公开(公告)日：2025-03-13

    申请号：US18465034

    申请日：2023-09-11

    Applicant: QUALCOMM Incorporated

    Inventor： An CHEN ,  Vijaya Datta MAYYURI

    IPC: G06N3/045 ,  G06N3/0495

    Abstract: A processor-implemented method for quantization-aware federated training includes quantizing, by a server, a global model. The global model is quantized at multiple different quantization levels for each of one or more subnetwork models to generate one or more quantized subnetwork models. The one or more subnetwork models are assigned to one or more of multiple devices according to device processing capabilities. The server distributes to one or more of the multiple devices, a quantized subnetwork model. The server receives a model update from the one or more devices based on local data. The server generates an updated global model according to an aggregation function based on the model update from each of the one or more devices.

Patent Agency Ranking