Generative Adversarial Network for Classification and Aimpoint Selection of Unmanned Air Vehicles
Midshipman Researcher(s): 1/C John Gale
Adviser(s): Professor Tae Lim
Poster #5
Target classification is an operator intensive process. As UAVs become more prevalent, software used to track and classify UAVs will require speed and accuracy. An automated classification system powered by a Generative Adversarial Network (GAN) and machine learning would greatly reduce the operator’s effort and improve speed and accuracy of decision making. This research will determine the viability of using GAN generated synthetic images to train a UAV classification system. If successful, synthetic images could be generated of UAVs in any desired orientation, weather, and lighting, providing time efficient training data while reducing the need for real images in training.
