中科微点-全球专利检索

  1. Login
  2. Sign Up

Search Results

11 records (took 0.01 seconds)

    METHOD AND DEVICE FOR AUGMENTING TRAINING DATA BY COMBINING OBJECT AND BACKGROUND
    1.
    发明申请
    METHOD AND DEVICE FOR AUGMENTING TRAINING DATA BY COMBINING OBJECT AND BACKGROUND 有权

    公开(公告)号：US20220207294A1

    公开(公告)日：2022-06-30

    申请号：US17137178

    申请日：2020-12-29

    Applicant: Markany Inc.

    Inventor： So Won KIM ,  Sae Yun JEON ,  Seung Yeob CHAE

    IPC: G06K9/62 ,  G06T11/60 ,  G06K9/00 ,  G06N3/08

    Abstract: Disclosed a method for augmenting training data by combining an object and a background with each other. The method includes extracting an object image, wherein the object image is a machine learning target; determining a type of the object image; receiving a background image, wherein the background image comprises a plurality of different background regions; identifying a first background region and a second background region among the plurality of different background regions; and combining the object image with the first background region and the second background region to augment training data, wherein combining the object image with the first background region and the second background region includes randomly positioning an image of a first type object corresponding to the first background region into the first background region, and randomly positioning an image of a second type object corresponding to the second background region into the second background region.

    METHOD AND APPARATUS FOR RECONSTRUCTING CONTENT IMAGE DATA
    4.
    发明申请
    METHOD AND APPARATUS FOR RECONSTRUCTING CONTENT IMAGE DATA 审中-公开

    公开(公告)号：US20200051227A1

    公开(公告)日：2020-02-13

    申请号：US16533732

    申请日：2019-08-06

    Applicant: Markany Inc.,

    Inventor： Ji Seop MOON ,  Min Soo PARK ,  Dong Young LEE ,  So Won KIM ,  Tai Yoon LEE

    IPC: G06T5/50 ,  G06T3/60 ,  G06T3/40 ,  G06T3/00 ,  G06T1/00

    Abstract: Provided is a method for reconstructing content image data. The method includes selecting a first point and a second point in a first image of first content, selecting a third point and a fourth point in a second image of second content (the second image is an image corresponding to the first image and the third point and the fourth point are points in an image corresponding to the first point and the second point, respectively), generating a first reference vector using the first point and the second point, generating a second reference vector using the third point and the fourth point, calculating a rotation, scale, and transformation (RST) value from the first image to the second image using the first reference vector and the second reference vector; and reconstructing the second content using the calculated RST value.

    WEIGHT INITIALIZATION METHOD AND APPARATUS FOR STABLE LEARNING OF DEEP LEARNING MODEL USING ACTIVATION FUNCTION
    5.
    发明申请
    WEIGHT INITIALIZATION METHOD AND APPARATUS FOR STABLE LEARNING OF DEEP LEARNING MODEL USING ACTIVATION FUNCTION 有权

    公开(公告)号：US20210201153A1

    公开(公告)日：2021-07-01

    申请号：US16729506

    申请日：2019-12-30

    Applicant: Markany Inc.,

    Inventor： Seung Yeob CHAE ,  So Won KIM ,  Min Soo PARK

    IPC: G06N3/08 ,  G06N3/04

    Abstract: Provided is an artificial neural network learning apparatus for deep learning. The apparatus includes an input unit configured to acquire an input data or a training data, a memory configured to store the input data, the training data, and a deep learning artificial neural network model, and a processor configured to perform computation based on the artificial neural network model, in which the processor sets the initial weight depending on the number of nodes belonging to a first layer and the number of nodes belonging to a second layer of the artificial neural network model, and determines the initial weight by compensation by multiplying a standard deviation (σ) by a square root of a reciprocal of a probability of a normal probability distribution for a remaining section except for a section in which an output value of the activation function converges to a specific value.

    METHOD AND APPARATUS FOR WATERMARKING OF DIGITAL CONTENT, METHOD FOR EXTRACTING INFORMATION
    6.
    发明申请
    METHOD AND APPARATUS FOR WATERMARKING OF DIGITAL CONTENT, METHOD FOR EXTRACTING INFORMATION 审中-公开

    公开(公告)号：US20190294761A1

    公开(公告)日：2019-09-26

    申请号：US15925774

    申请日：2018-03-20

    Applicant: MARKANY INC.

    Inventor： So Won KIM ,  Ji Seop MOON ,  Min Soo PARK ,  Hyungyu KIM ,  Euee Seon JANG

    IPC: G06F21/16 ,  H04L9/08

    Abstract: A method and apparatus for watermarking digital content and a data extraction method are disclosed. The digital content watermarking method includes: generating private and public keys corresponding to the user terminal; generating watermark data containing user identity information corresponding to the user terminal and user details associated with the watermark data; generating encrypted data by encrypting the watermark data and the user details with the private key; transmitting the watermark data, the encrypted data, and the public key to a server; receiving digital content when authorized by the server; and watermarking the digital content with the watermark data.

    WATERMARK EMBEDDING METHOD AND APPARATUS, AND SYSTEM
    8.
    发明申请
    WATERMARK EMBEDDING METHOD AND APPARATUS, AND SYSTEM 审中-公开

    公开(公告)号：US20180211354A1

    公开(公告)日：2018-07-26

    申请号：US15434065

    申请日：2017-02-16

    Applicant: MARKANY INC.

    Inventor： So Won KIM ,  Tai Yoon LEE ,  Min Soo PARK ,  Ji Seop MOON ,  Dong Hwan SHIN

    IPC: G06T1/00 ,  H04N21/8358

    CPC classification number: H04N21/8358 ,  G06T1/0028 ,  G06T2201/0051 ,  G06T2201/0202

    Abstract: Disclosed is a watermark embedding method. The method includes receiving an original image, generating a watermark image based on the selected watermark pattern, generating an alpha blended stego image by alpha blending the original image and the watermark image, and providing the alpha blended stego image, in which the watermark image is generated by considering at least one of allowable pixel value range and alpha value α range of the watermark pattern applied to a first pixel of the original image, wherein the at least one of allowable pixel value range and alpha value α range are determined by the difference of pixel value of the first pixel between the original image and the stego image by embedding the watermark pattern to the first pixel.

    METHOD FOR EXECUTING ACTIVATION FUNCTION FOR DEEP LEARNING ALGORITHM, AND APPARATUS FOR EXECUTING SAID METHOD
    9.
    发明申请
    METHOD FOR EXECUTING ACTIVATION FUNCTION FOR DEEP LEARNING ALGORITHM, AND APPARATUS FOR EXECUTING SAID METHOD 审中-公开

    公开(公告)号：US20200257981A1

    公开(公告)日：2020-08-13

    申请号：US16358701

    申请日：2019-03-20

    Applicant: MARKANY INC.

    Inventor： Seung Yeob CHAE ,  So Won KIM ,  Min Soo PARK

    IPC: G06N3/08 ,  G06N3/04 ,  G06F17/11

    Abstract: Disclosed is a method for executing an activation function for a deep learning algorithm. The method includes: determining whether an input value to a first node of an artificial neural network related to the deep learning algorithm is positive or negative; executing a first activation function in response to the input value being positive, or executing a second activation function in response to the input value being negative; and providing a value resulted from the execution of the first activation function or the second activation value to a second node of the artificial neural network, wherein the first activation function is a Rectified Linear Unit (ReLU) function, wherein the second activation function is a linear function having a first gradient in a first section of a negative number region and a second gradient in a second section of the negative number region.

Patent Agency Ranking