Simulation, GAN insights

We offer an additional ‘style transfer’ (realism improvement) service to eliminate the realism gap. General Adversarial Networks (GANs) are a means to improve the realism of synthetic images. The generator runs multiple iterations of optimization on synthetic images, until no differences can be detected while comparing them with real images.