知搜-全球专利检索

  1. 登录
  2. 注册
发明公开
CN110322418A 一种超分辨率图像生成对抗网络的训练方法及装置 审中-实审

一种超分辨率图像生成对抗网络的训练方法及装置
摘要：
本申请提供了一种超分辨率图像生成对抗网络的训练方法及装置，所述超分辨率图像生成对抗网络包括生成模型和损失计算模型，所述损失计算模型包括判别单元和分类单元，该方法包括：将图像信息输入到所述生成模型，以使所述生成模型输出超分辨率图像；将所述超分辨率图像输入到所述损失计算模型，以使所述损失计算模型计算所述超分辨率图像的总损失；根据所述超分辨率图像的总损失，对所述超分辨率图像生成对抗网络进行训练。
摘要（英）：

The invention provides a training method and device for a super-resolution image generative adversarial network. The super-resolution image generative adversarial network comprises a generation modeland a loss calculation model. The loss calculation model comprises a discrimination unit and a classification unit. The method comprises the steps: inputting image information into the generation model, so as to enable the generation model to output a super-resolution image; inputting the super-resolution image into the loss calculation model to enable the loss calculation model to calculate the total loss of the super-resolution image; and training the super-resolution image generative adversarial network according to the total loss of the super-resolution image.

0/0