IMAGE TRANSLATION METHOD AND DEVICE IMAGE TRANSLATION MODEL TRAINING METHOD AND DEVICE

摘要

The present invention discloses a method and apparatus for image translation, a method and apparatus for training an image translation model, and relates to the fields of deep learning and image processing. Specific implementation means include: obtaining an image translation request, wherein the translation request includes an original image; down-sampling the original image to generate a reduced image corresponding to the original image; According to the reduced image, a transformation parameter corresponding to each pixel point of the pre-translated image corresponding to the original image, the mask image, and the original image is generated, wherein the size of the pre-translated image and the mask image is the same as the size of the original image. same; transform the original image according to the transform parameter to generate a transformed image; fusing the transform image, the pre-translated image and the mask image to generate the target translation image. Therefore, while reducing the computational amount of image translation, it is possible to ensure the image translation effect, and improve the clarity of the translated target translation image.