Machine learning system, domain conversion device, and machine learning method

摘要

PROBLEM TO BE SOLVED: To efficiently learn a model for performing domain conversion of an image. A machine learning system includes a first determination model for determining whether or not an input image is a second domain image, and an extracted image obtained by extracting a region where an object is present from the input image from the second domain image. And a second determination model for determining whether or not the extracted region is an extracted image, and selecting either the pseudo second domain image or the second domain image and inputting it to the first determination model, Either the first extracted image of the second domain image or the second extracted image of the second domain image is selected and input to the image extraction unit, and the first determination model is learned based on the first determination result of the first determination model. The second determination model is learned based on the second determination result of the second determination model, and the first domain image obtained by photographing the target object is pseudo second domain based on the first determination result and the second determination result. A pseudo image generation model for converting into an image is learned. [Selection diagram] Figure 4