Accelerating Model Validation by Reconstructing Image Sets Using GAN

In recent years, the field of image recognition has been advancing rapidly, and various methods have been developed. However, it is necessary to perform many validations because the appropriate network structure differs depending on the type of image. In many examples, models have been validated using all the original image dataset, which requires a lot of effort. In this study, we used CGAN to reconstruct a small image dataset and succeeded in significantly reducing the model validation time.