首页> 外国专利> A method and apparatus for providing an integrated feature map using a plurality of output ensembles from convolution neural networksNtexored feature map using enamble of multi outcomes from consensus neural network

A method and apparatus for providing an integrated feature map using a plurality of output ensembles from convolution neural networksNtexored feature map using enamble of multi outcomes from consensus neural network

机译:一种方法和装置,用于使用来自卷积神经网络中的多个输出集成的集成功能映射使用来自互连神经网络的多结果的典型来看

摘要

To provide a method and a device for providing an integrated feature map by using an ensemble of a plurality of outputs from a convolutional neural network.SOLUTION: A method for acquiring an integrated feature map includes steps of: a CNN device receiving an input image and applying a plurality of deformation functions to deform the input image to generate a plurality of deformed input images; the CNN device applying convolution operation to each of the deformed input images to obtain each of deformed feature maps corresponding to each of the deformed input images; the CNN device applying each of reverse transform functions corresponding to each of the deformed functions, to each of the deformed feature maps, to generate each of reverse transform feature maps corresponding to each of the deformed feature maps; and the CNN device integrating at least a part of the reverse transform feature maps to obtain an integrated feature map.SELECTED DRAWING: Figure 1
机译:提供一种方法和设备,用于通过使用来自卷积神经网络的多个输出的集成的集成特征映射来提供集成特征映射。解决集成特征映射的方法包括以下步骤:接收输入图像的CNN设备应用多个变形函数以使输入图像变形以产生多个变形的输入图像;将卷积操作应用于每个变形的输入图像的CNN设备,以获得对应于每个变形的输入图像的变形特征映射;将对应于每个变形功能的每个反向变换功能应用于每个变形特征映射的CNN设备,以生成与每个变形的特征图相对应的反向变换特征映射;并且CNN设备集成了至少一部分反转变换特征映射以获取集成的特征映射。选择图:图1

著录项

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