中科微点-全球专利检索

  1. Login
  2. Sign Up

Search Results

12 records (took 0.019 seconds)

    Training Text Recognition Systems
    1.
    发明申请
    Training Text Recognition Systems 审中-公开

    公开(公告)号：US20200151503A1

    公开(公告)日：2020-05-14

    申请号：US16184779

    申请日：2018-11-08

    Applicant: Adobe Inc.

    Inventor： Zhaowen Wang ,  Hailin Jin ,  Yang Liu

    IPC: G06K9/62 ,  G06K9/34 ,  G06K9/32

    Abstract: In implementations of recognizing text in images, text recognition systems are trained using noisy images that have nuisance factors applied, and corresponding clean images (e.g., without nuisance factors). Clean images serve as supervision at both feature and pixel levels, so that text recognition systems are trained to be feature invariant (e.g., by requiring features extracted from a noisy image to match features extracted from a clean image), and feature complete (e.g., by requiring that features extracted from a noisy image be sufficient to generate a clean image). Accordingly, text recognition systems generalize to text not included in training images, and are robust to nuisance factors. Furthermore, since clean images are provided as supervision at feature and pixel levels, training requires fewer training images than text recognition systems that are not trained with a supervisory clean image, thus saving time and resources.

    Font recognition using adversarial neural network training
    2.
    发明授权
    Font recognition using adversarial neural network training 有权

    公开(公告)号：US10592787B2

    公开(公告)日：2020-03-17

    申请号：US15807028

    申请日：2017-11-08

    Applicant: Adobe Inc.

    Inventor： Yang Liu ,  Zhaowen Wang ,  Hailin Jin

    IPC: G06K9/68 ,  G06K9/62 ,  G06K9/00 ,  G06N3/04 ,  G06N3/08 ,  G06K9/66

    Abstract: The present disclosure relates to a font recognition system that employs a multi-task learning framework and adversarial training to improve font classification and remove negative side effects caused by intra-class variances of glyph content. For example, in one or more embodiments, the font recognition system adversarial trains a font recognition neural network by minimizing font classification loss while at the same time maximizing glyph classification loss. By employing an adversarially trained font classification neural network, the font recognition system can improve overall font recognition by removing the negative side effects from diverse glyph content.

    Font recognition by dynamically weighting multiple deep learning neural networks
    3.
    发明授权
    Font recognition by dynamically weighting multiple deep learning neural networks 有权

    公开(公告)号：US10515296B2

    公开(公告)日：2019-12-24

    申请号：US15812548

    申请日：2017-11-14

    Applicant: Adobe Inc.

    Inventor： Yang Liu ,  Zhaowen Wang ,  I-Ming Pao ,  Hailin Jin

    IPC: G06K9/62 ,  G06K9/68 ,  G06N5/04 ,  G06N3/08

    Abstract: The present disclosure relates to a font recognition system that employs a multi-task learning framework and training to improve font classification and remove negative side effects caused by intra-class variances of glyph content. For example, in one or more embodiments, the font recognition system trains a hybrid font recognition neural network that includes two or more font recognition neural networks and a weight prediction neural network. The hybrid font recognition neural network determines and generates classification weights based on which font recognition neural network within the hybrid font recognition neural network is best suited to classify the font in an input text image. By employing a hybrid trained font classification neural network, the font recognition system can improve overall font recognition as well as remove the negative side effects from diverse glyph content.

    Generating three-dimensional representations for digital objects utilizing mesh-based thin volumes
    6.
    发明授权
    Generating three-dimensional representations for digital objects utilizing mesh-based thin volumes 有权

    公开(公告)号：US12254570B2

    公开(公告)日：2025-03-18

    申请号：US17661878

    申请日：2022-05-03

    Applicant: Adobe Inc.

    Inventor： Sai Bi ,  Yang Liu ,  Zexiang Xu ,  Fujun Luan ,  Kalyan Sunkavalli

    IPC: G06T17/20 ,  G06T13/20

    Abstract: The present disclosure relates to systems, methods, and non-transitory computer readable media that generate three-dimensional hybrid mesh-volumetric representations for digital objects. For instance, in one or more embodiments, the disclosed systems generate a mesh for a digital object from a plurality of digital images that portray the digital object using a multi-view stereo model. Additionally, the disclosed systems determine a set of sample points for a thin volume around the mesh. Using a neural network, the disclosed systems further generate a three-dimensional hybrid mesh-volumetric representation for the digital object utilizing the set of sample points for the thin volume and the mesh.

    Training Text Recognition Systems
    7.
    发明申请
    Training Text Recognition Systems 有权

    公开(公告)号：US20210241032A1

    公开(公告)日：2021-08-05

    申请号：US17240097

    申请日：2021-04-26

    Applicant: Adobe Inc.

    Inventor： Zhaowen Wang ,  Hailin Jin ,  Yang Liu

    IPC: G06K9/62 ,  G06K9/32 ,  G06K9/34

    Abstract: In implementations of recognizing text in images, text recognition systems are trained using noisy images that have nuisance factors applied, and corresponding clean images (e.g., without nuisance factors). Clean images serve as supervision at both feature and pixel levels, so that text recognition systems are trained to be feature invariant (e.g., by requiring features extracted from a noisy image to match features extracted from a clean image), and feature complete (e.g., by requiring that features extracted from a noisy image be sufficient to generate a clean image). Accordingly, text recognition systems generalize to text not included in training images, and are robust to nuisance factors. Furthermore, since clean images are provided as supervision at feature and pixel levels, training requires fewer training images than text recognition systems that are not trained with a supervisory clean image, thus saving time and resources.

    Training text recognition systems
    8.
    发明授权
    Training text recognition systems 有权

    公开(公告)号：US10997463B2

    公开(公告)日：2021-05-04

    申请号：US16184779

    申请日：2018-11-08

    Applicant: Adobe Inc.

    Inventor： Zhaowen Wang ,  Hailin Jin ,  Yang Liu

    IPC: G06K9/62 ,  G06K9/32 ,  G06K9/34

    Abstract: In implementations of recognizing text in images, text recognition systems are trained using noisy images that have nuisance factors applied, and corresponding clean images (e.g., without nuisance factors). Clean images serve as supervision at both feature and pixel levels, so that text recognition systems are trained to be feature invariant (e.g., by requiring features extracted from a noisy image to match features extracted from a clean image), and feature complete (e.g., by requiring that features extracted from a noisy image be sufficient to generate a clean image). Accordingly, text recognition systems generalize to text not included in training images, and are robust to nuisance factors. Furthermore, since clean images are provided as supervision at feature and pixel levels, training requires fewer training images than text recognition systems that are not trained with a supervisory clean image, thus saving time and resources.

    GENERATING THREE-DIMENSIONAL REPRESENTATIONS FOR DIGITAL OBJECTS UTILIZING MESH-BASED THIN VOLUMES
    9.
    发明公开
    GENERATING THREE-DIMENSIONAL REPRESENTATIONS FOR DIGITAL OBJECTS UTILIZING MESH-BASED THIN VOLUMES 审中-公开

    公开(公告)号：US20230360327A1

    公开(公告)日：2023-11-09

    申请号：US17661878

    申请日：2022-05-03

    Applicant: Adobe Inc.

    Inventor： Sai Bi ,  Yang Liu ,  Zexiang Xu ,  Fujun Luan ,  Kalyan Sunkavalli

    IPC: G06T17/20 ,  G06T13/20

    CPC classification number: G06T17/205 ,  G06T13/20 ,  G06T2210/21

    Abstract: The present disclosure relates to systems, methods, and non-transitory computer readable media that generate three-dimensional hybrid mesh-volumetric representations for digital objects. For instance, in one or more embodiments, the disclosed systems generate a mesh for a digital object from a plurality of digital images that portray the digital object using a multi-view stereo model. Additionally, the disclosed systems determine a set of sample points for a thin volume around the mesh. Using a neural network, the disclosed systems further generate a three-dimensional hybrid mesh-volumetric representation for the digital object utilizing the set of sample points for the thin volume and the mesh.

    Training text recognition systems
    10.
    发明授权
    Training text recognition systems 有权

    公开(公告)号：US11810374B2

    公开(公告)日：2023-11-07

    申请号：US17240097

    申请日：2021-04-26

    Applicant: Adobe Inc.

    Inventor： Zhaowen Wang ,  Hailin Jin ,  Yang Liu

    IPC: G06V20/62 ,  G06V30/148 ,  G06F18/214 ,  G06V10/764

    CPC classification number: G06V20/62 ,  G06F18/214 ,  G06V10/764 ,  G06V20/63 ,  G06V30/153 ,  G06V2201/01

    Abstract: In implementations of recognizing text in images, text recognition systems are trained using noisy images that have nuisance factors applied, and corresponding clean images (e.g., without nuisance factors). Clean images serve as supervision at both feature and pixel levels, so that text recognition systems are trained to be feature invariant (e.g., by requiring features extracted from a noisy image to match features extracted from a clean image), and feature complete (e.g., by requiring that features extracted from a noisy image be sufficient to generate a clean image). Accordingly, text recognition systems generalize to text not included in training images, and are robust to nuisance factors. Furthermore, since clean images are provided as supervision at feature and pixel levels, training requires fewer training images than text recognition systems that are not trained with a supervisory clean image, thus saving time and resources.

Patent Agency Ranking