Research on Military Aerial Image Generation Technology Using Generative Artificial Intelligence

For military AI drones to perform surveillance, reconnaissance, and target attack missions, it is essential to have the technology to recognize the battlefield environment. To this end, video information acquired from onboard cameras mounted on drones can be utilized, and the drone can interpret the video information by itself using AI technology. In order to implement deep neural network-based object detection technology for military drones, a vast amount of aerial images and metadata of objects in the battlefield environment must be collected in advance. In this paper, we study the technology of automatically generating military aerial images using generative artificial intelligence. To this end, we implemented a virtual battlefield environment to collect aerial image datasets taken by drones, trained generative artificial intelligence, and analyzed the results of automatically generating various military aerial images.