Image conversion method and device, storage medium and electronic equipment

Abstract

The invention discloses an image conversion method and apparatus, a storage medium and an electronic device. The method comprises the steps of obtaining a first image generated by an image generation model based on an original image; inputting the first image and a second image of a preset target domain into a style discrimination model to obtain a first matrix; inputting the first image and the original image into a content discrimination model to obtain a second matrix; according to the first matrix, determining a style loss value of the first image through a preset style loss function; determining a content loss value of the first image through a preset content loss function according to the second matrix; and adjusting parameters of the image generation model according to the content loss value and the style loss value. According to the method, the parameters of the image generation model can be fed back and adjusted according to the content loss value and the style loss value, the precision of the image generation model is improved, the image generation model does not need to be subjected to deeper training and recognition, and the training process and the image conversion efficiency are improved.

Description

technical field [0001] The invention relates to the field of image processing, in particular to an image conversion method, device, storage medium and electronic equipment. Background technique [0002] In computer vision, many tasks can be described as image-to-image conversion, for example, the conversion between black and white images and color images, the conversion between aerial images and digital maps, etc. This type of conversion task has rich application scenarios, such as image enhancement, autonomous driving, etc. [0003] Traditional image conversion methods require cumbersome steps such as modeling, adding textures, and adding lighting. The key to the image conversion method is the generation of the image. Due to the complexity of the image data, the generated image needs to meet the following two requirements: the content is consistent with the original image, and the style is consistent with the target domain image. To ensure accuracy, existing deep convolut...

AI Technical Summary

Problems solved by technology

To ensure accuracy, existing deep convolutional networks have deep layers, resulting in inefficiencies in the training process and image generation

Images