首页> 外国专利> Learning device, learning method, learning program, data manufacturing method, image processing device, image processing method, and image processing program

Learning device, learning method, learning program, data manufacturing method, image processing device, image processing method, and image processing program

机译:学习设备,学习方法,学习程序,数据制造方法,图像处理设备,图像处理方法和图像处理程序

摘要

PROBLEM TO BE SOLVED: To provide a learning device capable of learning network parameters of a convolutional neural network with high accuracy and easily. A learning device (103) obtains a first training image and a correct answer image having a positional deviation with respect to the first training image, and a first from the correct answer image. Network parameters of the convolutional neural network (CNN) based on the generator (103b) that generates the second training image with higher resolution than the training image, and the first training image, the second training image, and the correct answer image. It has a learning unit (103c) for learning. [Selection diagram] Fig. 1
机译:解决的问题:提供一种能够高精度且容易地学习卷积神经网络的网络参数的学习装置。学习设备(103)获得第一训练图像和相对于第一训练图像具有位置偏差的正确答案图像,并且从正确答案图像中获得第一答案图像。基于生成器(103b)的卷积神经网络(CNN)的网络参数,该生成器以比训练图像,第一训练图像,第二训练图像和正确答案图像更高的分辨率生成第二训练图像。它具有用于学习的学习单元(103c)。 [选择图]图1

著录项

  • 公开/公告号JP2020154669A

    专利类型

  • 公开/公告日2020-09-24

    原文格式PDF

  • 申请/专利权人 キヤノン株式会社;

    申请/专利号JP20190052127

  • 发明设计人 木村 良範;

    申请日2019-03-20

  • 分类号G06T7;G06N3/08;

  • 国家 JP

  • 入库时间 2022-08-21 11:37:05

相似文献

  • 专利
  • 外文文献
  • 中文文献
获取专利

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号