Updating agent position information

    公开(公告)号:US11875570B1

    公开(公告)日:2024-01-16

    申请号:US17936754

    申请日:2022-09-29

    CPC classification number: G06V20/52 G06F18/22 G06V10/56 G06V10/809

    Abstract: Described is a multiple-camera system and process for re-identifying an agent located in a materials handling facility based on anterior views of agents. An anterior view of a newly detected agent may be partitioned and color signatures generated for each partition. Likewise, stored anterior views of agents (candidate agents) that may potentially be the newly detected agent are partitioned and color signatures generated for each partition. Based on the color signatures, a similarity between the anterior view of the newly detected agent and the candidate agents is determined. The similarity may be used to either determine that the newly detected agent is one of the candidate agents or reduce the set of candidate agents that are considered during a manual review.

    Composite agent representation
    2.
    发明授权

    公开(公告)号:US10692231B1

    公开(公告)日:2020-06-23

    申请号:US16541057

    申请日:2019-08-14

    Abstract: Described are systems and methods for determining an agent that performed an event within a materials handling facility. When an event is detected, an overhead image and a side-view image corresponding to the event location are obtained. Depth information is determined for the overhead image and a depth pattern generated that represents an agent. The depth pattern from the overhead image is overlaid onto the side-view image and the side-view image is segmented so that a representation of the agent included in the side-view image is extracted. A distance between the extracted representation of the agent and the event location is then determined. An association between the agent and the event may be determined based on the distance between the extracted representation and the event location.

    Associating an agent with an event based on multiple inputs

    公开(公告)号:US10388019B1

    公开(公告)日:2019-08-20

    申请号:US14747833

    申请日:2015-06-23

    Abstract: Described are systems and methods for determining an agent that performed an event within a materials handling facility. When an event is detected, an overhead image and a side-view image corresponding to the event location are obtained. Depth information is determined for the overhead image and a depth pattern generated that represents an agent. The depth pattern from the overhead image is overlaid onto the side-view image and the side-view image is segmented so that a representation of the agent included in the side-view image is extracted. A distance between the extracted representation of the agent and the event location is then determined. An association between the agent and the event may be determined based on the distance between the extracted representation and the event location.

    Agent re-identification
    4.
    发明授权

    公开(公告)号:US10592742B1

    公开(公告)日:2020-03-17

    申请号:US14868017

    申请日:2015-09-28

    Abstract: Described is a multiple-camera system and process for re-identifying an agent located in a materials handling facility based on anterior views of agents. An anterior view of a newly detected agent may be partitioned and color signatures generated for each partition. Likewise, stored anterior views of agents (candidate agents) that may potentially be the newly detected agent are partitioned and color signatures generated for each partition. Based on the color signatures, a similarity between the anterior view of the newly detected agent and the candidate agents is determined. The similarity may be used to either determine that the newly detected agent is one of the candidate agents or reduce the set of candidate agents that are considered during a manual review.

    Associating an event with an agent

    公开(公告)号:US12236715B1

    公开(公告)日:2025-02-25

    申请号:US18302723

    申请日:2023-04-18

    Inventor: Gang Hua

    Abstract: Described are systems and methods for determining an agent that performed an event within a materials handling facility. A series of overhead images that include representations of the event location and one or more agents are processed to determine a motion or movement of the agent over a period of time. For example, a motion model representative of a motion of the agent over a period of time is generated from the images. A distance between the motion model and the event location is also determined. An association between the agent and the event may be determined based on the motion model and the distance between the motion model and the event location.

    System to search three-dimensional object data

    公开(公告)号:US10902280B1

    公开(公告)日:2021-01-26

    申请号:US16460948

    申请日:2019-07-02

    Abstract: The presence of an object in a query image may be searched for using a feature of the query image. A local descriptor value for the feature is first determined. Previously stored object data for various 3D objects is then retrieved, with each set of stored object data including a probability distribution that describes a feature of a 3D object as obtained from images at different camera poses. A probability value is then determined using the local descriptor value for the feature in the query image and the available distribution data from the stored object data. The distribution data is then ranked based on the resulting probability values and the highest ranked distribution data is associated with the feature in the query image. Output data may also be generated to identify the 3D object in the query image based on an object identifier of the highest ranked distribution data.

    System to represent three-dimensional objects

    公开(公告)号:US10339411B1

    公开(公告)日:2019-07-02

    申请号:US14868205

    申请日:2015-09-28

    Inventor: Gang Hua Li Guan

    Abstract: A three-dimensional (3D) object may be represented by object data. The object data includes probability distributions that describe particular features on the 3D object. A probability distribution represents the value of local descriptors for the same feature as obtained from a plurality of images at different camera poses relative to the 3D object. A query image may be processed to determine features, and the query local descriptors of those features may be used to test against previously stored probability distributions. A combination of probability distributions that show a high probability of corresponding to the query local descriptors may be used to identify the 3D object appearing in the query image.

    Image partitioning for re-identification

    公开(公告)号:US11462005B1

    公开(公告)日:2022-10-04

    申请号:US16820298

    申请日:2020-03-16

    Abstract: Described is a multiple-camera system and process for re-identifying an agent located in a materials handling facility based on anterior views of agents. An anterior view of a newly detected agent may be partitioned and color signatures generated for each partition. Likewise, stored anterior views of agents (candidate agents) that may potentially be the newly detected agent are partitioned and color signatures generated for each partition. Based on the color signatures, a similarity between the anterior view of the newly detected agent and the candidate agents is determined. The similarity may be used to either determine that the newly detected agent is one of the candidate agents or reduce the set of candidate agents that are considered during a manual review.

    Associating an agent with an event using motion analysis

    公开(公告)号:US10891736B1

    公开(公告)日:2021-01-12

    申请号:US14747846

    申请日:2015-06-23

    Inventor: Gang Hua

    Abstract: Described are systems and methods for determining an agent that performed an event within a materials handling facility. A series of overhead images that include representations of the event location and one or more agents are processed to determine a motion or movement of the agent over a period of time. For example, a motion model representative of a motion of the agent over a period of time is generated from the images. A distance between the motion model and the event location is also determined. An association between the agent and the event may be determined based on the motion model and the distance between the motion model and the event location.

Patent Agency Ranking