利索-全球专利检索

  1. 登录
  2. 注册

搜索结果

1022 条记录 (耗时 0.031 秒)

    UTILIZING INTERACTIVE DEEP LEARNING TO SELECT OBJECTS IN DIGITAL VISUAL MEDIA
    1.
    发明申请
    UTILIZING INTERACTIVE DEEP LEARNING TO SELECT OBJECTS IN DIGITAL VISUAL MEDIA 审中-公开

    公开(公告)号：US20190236394A1

    公开(公告)日：2019-08-01

    申请号：US16376704

    申请日：2019-04-05

    申请人： Adobe Inc.

    发明人： Brian Price ,  Scott Cohen ,  Mai Long ,  Jun Hao Liew

    IPC分类号： G06K9/32 ,  G06K9/46

    CPC分类号： G06K9/3241 ,  G06K9/4628 ,  G06K2009/366

    摘要： Systems and methods are disclosed for selecting target objects within digital images utilizing a multi-modal object selection neural network trained to accommodate multiple input modalities. In particular, in one or more embodiments, the disclosed systems and methods generate a trained neural network based on training digital images and training indicators corresponding to various input modalities. Moreover, one or more embodiments of the disclosed systems and methods utilize a trained neural network and iterative user inputs corresponding to different input modalities to select target objects in digital images. Specifically, the disclosed systems and methods can transform user inputs into distance maps that can be utilized in conjunction with color channels and a trained neural network to identify pixels that reflect the target object.

    SYSTEMS AND METHODS FOR HARDWARE-BASED POOLING
    4.
    发明申请
    SYSTEMS AND METHODS FOR HARDWARE-BASED POOLING 审中-公开

    公开(公告)号：US20190205738A1

    公开(公告)日：2019-07-04

    申请号：US15862369

    申请日：2018-01-04

    申请人： Tesla, Inc.

    发明人： Peter Joseph BANNON ,  Kevin Altair Hurd

    IPC分类号： G06N3/063 ,  G06K9/66 ,  G06F15/18 ,  G06K9/34 ,  G06K9/46

    CPC分类号： G06N3/063 ,  G06K9/342 ,  G06K9/4628 ,  G06K9/66 ,  G06N20/00

    摘要： Described herein are systems and methods that utilize a novel hardware-based pooling architecture to process the output of a convolution engine representing an output channel of a convolution layer in a convolutional neural network (CNN). The pooling system converts the output into a set of arrays and aligns them according to a pooling operation to generate a pooling result. In certain embodiments, this is accomplished by using an aligner that aligns, e.g., over a number of arithmetic cycles, an array of data in the output into rows and shifts the rows relative to each other. A pooler applies a pooling operation to a combination of a subset of data from each row to generate the pooling result.

    METHOD AND APPARATUS FOR DISTRIBUTED EDGE LEARNING
    7.
    发明申请
    METHOD AND APPARATUS FOR DISTRIBUTED EDGE LEARNING 审中-公开

    公开(公告)号：US20190164020A1

    公开(公告)日：2019-05-30

    申请号：US15824389

    申请日：2017-11-28

    申请人： MOTOROLA SOLUTIONS, INC.

    发明人： Shervin Sabripour

    IPC分类号： G06K9/66 ,  G06N99/00 ,  G06K9/46 ,  G06K9/00

    CPC分类号： G06K9/66 ,  G06K9/00228 ,  G06K9/00369 ,  G06K9/00664 ,  G06K9/4628 ,  G06K9/6267 ,  G06N20/00 ,  H04N5/2258 ,  H04N5/232

    摘要： A portable electronic device and method. The portable electronic device includes a first camera, a second camera, an electronic processor, and one or more sensors. The electronic processor is configured to detect, based on information obtained from the one or more sensors, an incident and select a camera responsive to the incident. The electronic processor is further configured to capture an image using the selected camera and determine, within the image, a subject of interest, wherein the subject of interest is at least one selected from the group consisting of a person, an object, and an entity. The electronic processor is also configured to initiate an edge learning process on the subject of interest to create a classifier for use in identifying the subject of interest and transmit the classifier to a second portable electronic device within a predetermined distance from the portable electronic device.

    Distance Metric Learning Using Proxies
    10.
    发明申请
    Distance Metric Learning Using Proxies 审中-公开

    公开(公告)号：US20190065899A1

    公开(公告)日：2019-02-28

    申请号：US15710377

    申请日：2017-09-20

    申请人： Google Inc.

    发明人： Yair Movshovitz-Attias ,  King Hong Leung ,  Saurabh Singh ,  Alexander Toshev ,  Sergey Ioffe

    IPC分类号： G06K9/62 ,  G06N99/00

    CPC分类号： G06K9/6215 ,  G06K9/4628 ,  G06K9/6232 ,  G06K9/6255 ,  G06K9/6256 ,  G06K9/6262 ,  G06K9/66 ,  G06N20/00

    摘要： The present disclosure provides systems and methods that enable distance metric learning using proxies. A machine-learned distance model can be trained in a proxy space in which a loss function compares an embedding provided for an anchor data point of a training dataset to a positive proxy and one or more negative proxies, where each of the positive proxy and the one or more negative proxies serve as a proxy for two or more data points included in the training dataset. Thus, each proxy can approximate a number of data points, enabling faster convergence. According to another aspect, the proxies of the proxy space can themselves be learned parameters, such that the proxies and the model are trained jointly. Thus, the present disclosure enables faster convergence (e.g., reduced training time). The present disclosure provides example experiments which demonstrate a new state of the art on several popular training datasets.