A Shared Representation for Photorealistic Driving Simulators
This work addresses the need for high-fidelity simulators in autonomous vehicle training, offering a generic improvement for conditional image synthesis, though it is incremental as it builds on existing cGAN models.
The authors tackled the problem of synthesizing photorealistic images from semantic inputs like segmentation maps or poses, by proposing a new semantically-aware discriminator that improves image quality across scene, building, and human synthesis tasks on three datasets.
A powerful simulator highly decreases the need for real-world tests when training and evaluating autonomous vehicles. Data-driven simulators flourished with the recent advancement of conditional Generative Adversarial Networks (cGANs), providing high-fidelity images. The main challenge is synthesizing photorealistic images while following given constraints. In this work, we propose to improve the quality of generated images by rethinking the discriminator architecture. The focus is on the class of problems where images are generated given semantic inputs, such as scene segmentation maps or human body poses. We build on successful cGAN models to propose a new semantically-aware discriminator that better guides the generator. We aim to learn a shared latent representation that encodes enough information to jointly do semantic segmentation, content reconstruction, along with a coarse-to-fine grained adversarial reasoning. The achieved improvements are generic and simple enough to be applied to any architecture of conditional image synthesis. We demonstrate the strength of our method on the scene, building, and human synthesis tasks across three different datasets. The code is available at https://github.com/vita-epfl/SemDisc.