Detection method using fusion network based on attention mechanism, and terminal device

    公开(公告)号:US11222217B1

    公开(公告)日:2022-01-11

    申请号:US17369990

    申请日:2021-07-08

    摘要: A lane line detection method using a fusion network based on an attention mechanism, and a terminal device are provided. The method includes: synchronously acquiring natural images and point cloud data of a road surface; and inputting the natural images and the point cloud data into a pre-built and trained fusion network to output a lane line detection result. Time series frames and an attention mechanism are added to the fusion network to perform information fusing on the point cloud data and the natural images. Specifically, continuous frames are used to improve detection network performance to deal with complex situations such as label loss and vehicle being blocked; low-dimensional features are stitched with high-dimensional features by Skip Connection to make up for image detail information that is continuously lost as the network goes deeper, and the Decoder module is used to restore image dimensions to obtain a final result.

    TECHNIQUE FOR GENERATING A LABELED SET OF IMAGES

    公开(公告)号:US20210406580A1

    公开(公告)日:2021-12-30

    申请号:US17356720

    申请日:2021-06-24

    摘要: A method for generating a labeled set of images for use in machine learning based stray light characterization for space-related optical systems. The method comprises (a) obtaining a set of images simulated for a space-related optical system, wherein the images of the set of images contain stray light simulated for the space-related optical system, (b) for each image of the set of images, identifying one or more clusters of light contained in the respective image and labeling the respective image by the one or more clusters of light, wherein the one or more clusters of light comprise at least one cluster of stray light, and (c) creating, based on the labeled images of the set of images, a plurality of new labeled images by applying transformations to the labeled images to generate an augmented set of labeled images.

    Generating training data for object detection

    公开(公告)号:US11200455B2

    公开(公告)日:2021-12-14

    申请号:US16691746

    申请日:2019-11-22

    IPC分类号: G06K9/62 G06K9/46 G06K9/44

    摘要: A method for generating training data for object detection is provided. The method includes dividing an image into a plurality of sliced images. The image includes meta information for an object in the image, the meta information including bounding box information associated with a bounding box surrounding the object, and marker information for markers associated with the object, the markers being located within the bounding box. The method includes designating at least one sliced object image from among the plurality of sliced images, at least a portion of the bounding box overlapping with each of the respective sliced object images. The method also includes designating at least one training data image from among the designated sliced object images based on the marker information. The method also includes generating the training data from the designated training data images.

    IMAGE PROCESSING DEVICE, CONTROL METHOD, AND CONTROL PROGRAM

    公开(公告)号:US20210192694A1

    公开(公告)日:2021-06-24

    申请号:US17263404

    申请日:2018-10-29

    申请人: PFU LIMITED

    IPC分类号: G06T5/00 G06K9/44 G06K9/46

    摘要: Provided are an image processing apparatus, a control method, and a control program to more accurately remove the shading from the image including the shading. An image processing apparatus includes a storage device to store a shading pattern in which a condition of gradation values of a target pixel and a plurality of pixels having a predetermined positional relationship with respect to the target pixel are set, an acquisition module to acquire a multiple value image, a binary image generation module to generate a binary image from the multiple value image, a detection module to detect a pixel satisfying the condition set in the shading pattern from the multiple value image as a part of a shading, a shading removal pattern generation module to generate a shading removal pattern for removing the detected shading, based on the multiple value image, a shading removal image generation module to generate a shading removal image by applying the shading removal pattern to the binary image, and an output device to output the shading removal image or information generated using the shading removal image.

    LANE LINE CREATION FOR HIGH DEFINITION MAPS FOR AUTONOMOUS VEHICLES

    公开(公告)号:US20210172756A1

    公开(公告)日:2021-06-10

    申请号:US17115576

    申请日:2020-12-08

    申请人: DeepMap Inc.

    摘要: An HD map system represents landmarks on a high definition map for autonomous vehicle navigation, including describing spatial location of lanes of a road and semantic information about each lane, and along with traffic signs and landmarks. The system generates lane lines designating lanes of roads based on, for example, mapping of camera image pixels with high probability of being on lane lines into a three-dimensional space, and locating/connecting center lines of the lane lines. The system builds a large connected network of lane elements and their connections as a lane element graph. The system also represents traffic signs based on camera images and detection and ranging sensor depth maps. These landmarks are used in building a high definition map that allows autonomous vehicles to safely navigate through their environments.

    Lane line creation for high definition maps for autonomous vehicles

    公开(公告)号:US10859395B2

    公开(公告)日:2020-12-08

    申请号:US15859194

    申请日:2017-12-29

    申请人: DeepMap Inc.

    摘要: An HD map system represents landmarks on a high definition map for autonomous vehicle navigation, including describing spatial location of lanes of a road and semantic information about each lane, and along with traffic signs and landmarks. The system generates lane lines designating lanes of roads based on, for example, mapping of camera image pixels with high probability of being on lane lines into a three-dimensional space, and locating/connecting center lines of the lane lines. The system builds a large connected network of lane elements and their connections as a lane element graph. The system also represents traffic signs based on camera images and detection and ranging sensor depth maps. These landmarks are used in building a high definition map that allows autonomous vehicles to safely navigate through their environments.

    Article Recognition System and Article Recognition Method

    公开(公告)号:US20200342224A1

    公开(公告)日:2020-10-29

    申请号:US16851647

    申请日:2020-04-17

    申请人: Hitachi, Ltd.

    摘要: A technology capable of dealing with a variation in appropriate recognition parameters due to a change in work environment at a work site at a time of recognizing an article in an image. An article recognition system includes: an input section to which a captured image is input; an output section that presents information to a user; a storage section that stores master data defining features of an object to be recognized; a recognition section that detects the object to be recognized in the image; and a recognition evaluation section that evaluates a recognition content of the recognition section and that presents an alert to the output section on the basis of an evaluation result.