With recent progress in graphics, it has become more tractable to train models on synthetic images, potentially avoiding the need for expensive annotations. However, learning from synthetic images may not achieve the desired performance due to a gap between synthetic and real image distributions. To reduce this gap, we propose Simulated+Unsupervised (S+U) learning, where the task is to learn a model to improve the realism of a simulator's output using unlabeled real data, while preserving the annotation information from the simulator. We develop a method for S+U learning that uses an adversarial network similar to Generative Adversarial Networks (GANs), but with synthetic images as inputs instead of random vectors. We make several key modifications to the standard GAN algorithm to preserve annotations, avoid artifacts, and stabilize training: (i) a 'self-regularization' term, (ii) a local adversarial loss, and (iii) updating the discriminator using a history of refined images. We show that this enables generation of highly realistic images, which we demonstrate both qualitatively and with a user study. We quantitatively evaluate the generated images by training models for gaze estimation and hand pose estimation. We show a significant improvement over using synthetic images, and achieve state-of-the-art results on the MPIIGaze dataset without any labeled real data.

No abstract

No abstract

No abstract

To facilitate further research in emergent turn-taking, we propose a metric for evaluating the extent to which agents take turns using a shared resource. Our measure reports a turn-taking value for a particular time and a particular time scale, or 'resolution,' in a way that matches intuition. We describe how to evaluate the results of simulations where turn-taking may or may not be present and analyse the apparent turn-taking that could be observed between random independent agents. We illustrate the use of our turn-taking metric by reinterpreting previous work on turn-taking in emergent communication and by analysing a recorded human conversation.