Bidirectional image conversion system and method based on deep learning

Abstract

The invention provides a bidirectional image conversion system based on deep learning. The system comprises a bidirectional generator which can respectively perform an image conversion task between apair of image domains in the positive direction and the negative direction of the bidirectional generator, and in any direction of the bidirectional generator, a model calculates a target conversion image of an image by using a depth parallel computing framework for input multi-channel image data; and a discriminator which is used for carrying out quality evaluation on the image obtained by the bidirectional generator and the real image, and a quality evaluation result is used for training the bidirectional generator and the discriminator. Meanwhile, the invention provides a bidirectional image conversion method based on system implementation. According to the system, a bidirectional generator structure is provided, parameters of a deep learning model are greatly reduced on the premise that the image generation quality is not reduced, and two pairs of image conversion tasks can be carried out in one model at the same time under the supervised condition.

Description

technical field [0001] The present invention relates to the technical field of image conversion, in particular to a two-way image conversion system and method based on deep learning. The system and method are based on minimum mean square error reconstruction and are a method against generative networks and deep parallel computing frameworks. Background technique [0002] Image conversion refers to the task of image conversion between two image domains, which converts an image from image A into an image belonging to image B according to certain rules or requirements. The result of the conversion may be to keep the content of image domain A unchanged, but introduce the characteristics of image domain B, such as style transfer tasks, or it may be generated by the neural network based on less information in image domain A to contain more information (such as color information ) of image domain B, such as colorization (colorization) tasks, it is also possible to change the conten...

AI Technical Summary

Problems solved by technology

[0007] However, the existing image conversion technology still has the problem of a large number of model parameters and a single conversion task, which cannot really meet the needs of image conversion.

At present, there is no description or report of the similar technology of the present invention, and no similar data at home and abroad have been collected yet.

Images