2022 年 59 巻 8 号 p. 394-399
In recent years, data “generation” techniques using generative adversarial networks (GANs), in which multiple neural networks learn in an adversarial manner, have achieved remarkable results in many fields. The authors are participating in a project commissioned by the Ministry of Agriculture, Forestry and Fisheries (MAFF) called the “Artificial Intelligence Future Agriculture Creation Project” and are working on the development of automatic diagnosis of plants based on images and related technologies. In this paper, we introduce the trend of automatic diagnosis technology for plant diseases, the unique difficulties of this task, and the remarkable effect of GAN’s image generation technology on it.